Mining Patterns in Genomic and Clinical Cancer Data to Characterize Novel Driver Genes

Mining patterns in genomic and clinical cancer data to characterize novel driver genes

Rachel D. Melamed

Submitted in partial fulfillment of the

requirements for the degree

of Doctor of Philosophy

under the Executive Committee

of the Graduate School of Arts and Sciences

COLUMBIA UNIVERSITY

2015

Rachel D. Melamed

ABSTRACT

Mining patterns in genomic and clinical cancer data to characterize novel driver genes

Rachel D. Melamed

Cancer research, like many areas of science, is adapting to a new era characterized by increasing quantity, quality, and diversity of observational data. An example of the advances, and the resulting challenges, is represented by The Cancer Genome Atlas, an enormous public effort that has provided genomic profiles of hundreds of tumors of each of the most common solid cancer types. Alongside this resource is a host of other data and knowledge, including gene interaction databases, Mendelian disease causal variants, and electronic health records spanning many millions of patients. Thus, a current challenge is how best to integrate these data to discover mechanisms of oncogenesis and cancer progression. Ultimately, this could enable genomics- based prediction of an individual patient’s outcome and targeted therapies, a goal termed precision medicine. In this thesis, I develop novel approaches that examine patterns in populations of cancer patients to identify key genetic changes and suggest likely roles of these driver genes in the diseases.

In the first section I show how genomics can lead to the identification of driver alterations in melanoma. The most recurrent genetic mutations are often in important cancer driver genes: in a newly sequenced melanoma cohort, recurrent inactivating mutations point to an exciting new melanoma candidate tumor suppressor, FBXW7, with therapeutic implications.

But each tumor is unique, underlining the fact that recurrence will never capture all relevant mutations responsible for the disease. Tumors are a result of random events that must collaborate

to endow a cell with all of the invasive and immortal properties of a cancer. Some combinations of events are lethal to a developing tumor, while other combinations are simply not preferentially selected. In order to discover these complex patterns, I develop a method based on the joint entropy of a set of genes, called GAMToC. Using GAMToC, I identify sets of recurrently altered genes with a strongly non-random joint pattern of co-occurrence and mutual exclusivity. Then, I extend this method as a means of identifying novel genes with a role in cancer, by virtue of their non-random pattern of alteration. Insights into the roles of these novel drivers can come from their most strongly co-selected partners.

In the final section of the main text, I develop the use of cancer comorbidity, or increased cancer risk, as a novel data source for understanding cancer. The recent availability of clinical records spanning a large percentage of the American population has enabled discovery of many cancer comorbidities. Although most cancers arise as a result of somatic mutations accumulating over a patient’s lifespan, mutations present at birth could predispose some rare populations to increased cancer risk. Mendelian disease phenotype provides strong insight into the genotype of an afflicted individual. Thus, if Mendelian diseases with cancer comorbidity can be shown to have specific defects in processes that are important in the development of that cancer, statistical comorbidity could provide a new a resource for prioritizing Mendelian disease genes as novel cancer related genes. For this purpose, I integrate clinical comorbidity, Mendelian disease causal variants, and somatic genomic profiles of thousands of cancers. I demonstrate that comorbidity indeed is associated with significant genetic similarity between Mendelian diseases and the cancers these patients are predisposed to, suggesting highly interesting and plausible new candidate cancer genes. While cancer may be the result of a series of selected random events, patterns of incidence across large populations, as measured by genomics or by other phenotypes, contain much non- random signal yet to be mined.

TABLE OF CONTENTS

LIST OF GRAPHS, IMAGES, AND ILLUSTRATIONS ...... iv

LIST OF SUPPLEMENTARY TABLES ...... vi

ACKNOWLEDGEMENTS ...... vii

1 INTRODUCTION ...... 1

2 Coding mutations influencing development of melanoma and nevi ...... 9

2.1 Sequencing melanomas and discovery of FBXW7 as a melanoma tumor suppressor ...... 9

2.1.1 Methods ...... 10

2.1.2 Results ...... 11

2.2 Sequencing nevi: exploring the progression to melanoma ...... 13

2.2.1 Methods ...... 14

2.2.2 Results and discussion ...... 15

2.3 Discussion ...... 20

3 Applying the total correlation to identify and contextualize driver alterations ...... 22

3.1 An information theoretic method to identify combinations of genomic alterations that

promote glioblastoma ...... 23

3.1.1 Introduction ...... 23

3.1.2 Method ...... 27

3.1.3 Results ...... 33

3.1.4 Discussion ...... 43

3.2 GAMToC-L: Using patterns of co-selection of cancer genes to identify and contextualize

novel drivers ...... 48

3.2.1 Methods ...... 50

3.2.2 Results ...... 56

3.2.3 Discussion ...... 62

4 Genetic similarity between cancers and comorbid Mendelian diseases identifies candidate driver genes ...... 64

4.1 Introduction ...... 65

4.2 Comparing Mendelian disease and comorbid cancer ...... 67

4.2.1 Integration of disease comorbidities and genes ...... 67

4.2.2 Genetic similarity of comorbid diseases ...... 71

4.3 Mendelian disease comorbidity and cancer processes ...... 78

4.3.1 Prediction of diseases with shared cellular processes ...... 78

4.3.2 Pan-cancer Mendelian associations ...... 87

4.4 Discussion ...... 91

5 Data-driven discovery of seasonally linked diseases from an Electronic Health Records system ...... 95

5.1 Introduction ...... 96

5.2 Methods ...... 99

5.2.1 Quantifying incidence of diagnoses ...... 99

5.2.2 Correcting for confounding trends ...... 99

5.2.3 Evaluating periodicity ...... 100

5.2.4 Comorbidity analysis ...... 101

5.3 Results ...... 102

5.3.1 LSP-detrend: finding periodic signal ...... 102

5.3.2 Major types of periodic signal and known seasonal disease ...... 104

5.3.3 Confirmation of recent reports of seasonal effects ...... 105

5.3.4 Novel findings: acute exacerbations of myasthenia gravis ...... 108

5.3.5 Dissecting causes of seasonality of acute exacerbations by finding comorbid diseases

109

5.3.6 Comparison between hospital systems ...... 110

5.4 Discussion ...... 111

6 CONCLUSION ...... 113

7 Supplementary Tables ...... 118

8 REFERENCES ...... 124

iii

LIST OF GRAPHS, IMAGES, AND ILLUSTRATIONS

Figure 1-1: Subtypes of glioblastoma...... 5

Figure 2-1: FBXW7 as a novel therapeutic target in melanoma ...... 13

Figure 2-2: Mutation spectrum of nevus types, and frequency thresholds derived per nevus ...... 16

Figure 3-1: Workflow of GAMToC gene set finding...... 26

Figure 3-2 Illustration of copy number linkages in the GBM cohort ...... 28

Figure 3-3: Visualization of the relationship between temperature (in legend), change in total

correlation (x-axis), and acceptance probability in the simulated annealing (y-axis)...... 32

Figure 3-4: Ability to find multi-gene co-mutational patterns...... 34

Figure 3-5: Comparison of different methods in GBM mutation data...... 36

Figure 3-6: Recovery of different module sizes in only mutation data...... 38

Figure 3-7: Networks of total correlation modules...... 39

Figure 3-8: Networks seeded with query genes...... 42

Figure 3-9 Cell cycle, DNA damage, and mitogenic gene subtype associations...... 47

Figure 3-10 Effect of decreasing temperature ...... 51

Figure 3-11 Distribution of the frequency of genes and gene pairs appearing in the module data.

...... 53

Figure 3-12 Frequency of co-selection of pairs of genes in the module data...... 54

Figure 3-13 GAMToC-L module for the GBM data...... 57

Figure 3-14: Second module from GBM data. For legend see Figure 3-13...... 60

Figure 3-15 Module for lower grade glioma. For legend see Figure 3-13...... 62

Figure 4-1 Distribution of number of genes per disease...... 69

Figure 4-2 Characteristics of Mendelian diseases ...... 70

Figure 4-3 Outline of the approach...... 71

Figure 4-4 Genes shared in comorbid diseases ...... 73

Figure 4-5 Aggregate similarity of comorbid diseases ...... 76

Figure 4-6 Depiction of comorbid diseases with skin melanoma ...... 79

Figure 4-7 Analysis of the role of albinism related genes in melanoma...... 81

Figure 4-8 Pairwise pathway metric for Rubinstein-Taybi and melanoma ...... 82

Figure 4-9 Coexpression of ectodermal dysplasia genes with PTK6 ...... 83

Figure 4-10 GSEA plot of the ectodermal dysplasia candidates ...... 84

Figure 4-11 Interaction of Diamond-Blackfan anemia genes with glioblastoma altered genes. .... 85

Figure 4-12 GSEA plot of holoprosencephaly candidate genes ...... 87

Figure 4-13 The distribution of the number of comorbid cancer diagnosis codes per Mendelian

disease ...... 88

Figure 4-14 Mendelian diseases with broad cancer links ...... 90

Figure 5-1: Identifying confounding factors in temporal diagnosis ...... 103

Figure 5-2: Pre-processed and row-normalized monthly incidence for 227 codes with periodic

signal...... 104

Figure 5-3: Selected diseases with periodic signal...... 107

Figure 5-4: Overall seasonality of hospitalization in Columbia and Stanford ...... 110

LIST OF SUPPLEMENTARY TABLES Supplementary Table 1: Significantly mutated genes in the melanoma cohort, and their mutations

across the tumors ...... 118

Supplementary Table 2: Genes significantly less frequently mutated in the nevus cohort, see

2.2.2.4 ...... 119

Supplementary Table 3: Pairs of comorbid and genetically similar Mendelian disease and cancer,

related to 4.3. Columns described below: ...... 120

Supplementary Table 4: Continuation of Supplementary Table 3 ...... 122

Supplementary Table 5: ADAMS results for comorbidity with acute exacerbations of myasthenia

gravis. Related to section 5.3.5...... 123

ACKNOWLEDGEMENTS I would like to acknowledge all who contributed to this dissertation. First, my advisor from my days before graduate school, Christophe Benoist, who encouraged me to pursue the PhD. I would like to thank my graduate advisor Raul Rabadan for his support in his lab, as well as all of my other co-authors on related work. In particular, Andrey Rzhetsky, Jiguang Wang, Antonio

Iavarone, Hossein Khiabanian, Julide Celebi, and Iraz Aydin contributed to work described in this dissertation. I am grateful to all members of my committee for supervising this work: Raul

Rabadan, Antonio Iavarone, George Hripcsak, Harmen Bussemaker, and Yufeng Shen.

Additionally, I would like to thank members of the Biomedical Informatics department administration for years of help. On a personal note I thank friends from graduate school for all of their moral support in the process, including Bo-Juen Chen, Denesy Mancenido, Felix Sanchez

Garcia, Regina Lutz, Francesco Abate and others. My closest friends and loved ones Tommy and

Samantha made it possible for me to continue during some difficult times, and I hope I can do the same for them.

vii

1 INTRODUCTION

Since the discovery of the first oncogenic genetic lesions, the hunt for causal mutations driving cancer has been driven by the promise of understanding the basic biology of tumors, forecasting patient outcome, and finding druggable targets. The Philadelphia chromosome, a chromosomal fusion discovered via cytogenetic analysis of leukemia patients in 1960 (Nowell and Hungerford

1960), leads to hyperactivation of the Abl tyrosine kinase and sustained, self-sufficient growth and proliferation signaling. Excitingly, the Philadelphia chromosome is found in 95% of chronic myelogenous leukemias, and the fusion protein can be inhibited with imatinib, with effective clinical response(Druker et al. 2001). Others suspected that this growth signaling pathway may harbor other common mutations that drive tumor growth, and sequenced the BRAF gene in a number of tumors, finding recurrent activating mutations at a single site, particularly in melanoma (Davies et al. 2002). This led to development of vemurafinib to target this mutation.

However, vemurafinib is known to fail by 18 months after treatment of melanoma patients(Poulikakos and Rosen 2011). Additionally, only around 60% of melanomas have BRAF mutation. Most cancers appear to have far more heterogeneity, and more complexity, than was initially hoped.

In fact, knowledge about the processes underlying cancer development can suggest explanations for the heterogeneity between tumors, and for the difficulties in treating some cancers. Cancer usually requires complementary alterations to multiple cellular functions. For example, the same

BRAF activating mutation found in many melanomas, or a similarly activating mutation of NRAS, is also found in most benign nevi. The acquisition of this mutation does lead to some proliferation in these cells, resulting in the nevus, but these growths are benign, self-limited by the phenomenon of oncogene induced senescence (Michaloglou et al. 2005).

The necessity to bypass this checkpoint represents just one obstacle the tumor must overcome. A number of changes are necessary in the tumor evolutionary process, including reduced susceptibility to growth inhibition signals, angiogenesis, ability to invade surrounding tissue, and genomic instability that can accelerate the evolutionary process (Hanahan and Weinberg 2011).

The necessity for multiple alterations in cancer development was supported decades ago(Armitage and Doll 1954). Armitage and Doll considered that advanced age of onset for most cancers may reflect the time it takes for multiple mutations to accumulate in a cell, precipitating transformation. Modeling the distribution of cancer onset age, they found it was consistent with the requirement for multiple successive mutations to occur. One rare young-onset cancer is retinoblastoma, which includes a familial form often involving multiple tumors over a lifetime.

Knudson modeled the variable number of tumors per patient, and age of onset, in familial retinoblastoma. The analysis suggested a “two-hit” hypothesis, where the germline mutation present in all cells of all carriers must be accompanied by a somatic mutation(Knudson 2001).

Even in relatively simple cancers with a strong inherited genetic component, random somatic mutations determine cancer development. Multiple somatic mutations are required for tumorigenesis in most common cancers.

The findings of recurrent genetic changes, such as BCR-Abl fusion and BRAF mutation, as well as the understanding that cancer is the result of heterogeneous combinations of somatic alterations, have encouraged the development of larger scale cancer genomics efforts. The International

Cancer Genome Consortium, and, in the USA, The Cancer Genome Atlas (TCGA), aim to profile most common cancers. These projects collect complex profiles including mutation, copy number, gene expression, and epigenetics, all with the goal of understanding how genes are mis-regulated

in cancer. Thus far, TCGA has unrestricted releases of copy number and whole exome sequencing data for thousands of patients across 19 cancers.

One of the principal goals of somatic genomic tumor collections continues to be to distinguish

“driver” genes from the large number of “passenger” mutations randomly mutated and passively retained. Drivers include oncogenes like BRAF that promote tumor growth, as well as tumor suppressors, genes that can inhibit growth and are often disabled in the cell. Prevalent current methods for identifying drivers look for the most recurrent genetic lesions across cancer genomics profiles, such as copy number or whole exome sequencing results. Copy number aberrations (CNA) can indicate deletions of a gene, that can disable tumor suppressor genes by lowering or abolishing their expression, as well as gene amplifications, which result in extra copies of a gene and overly expressed, and thus overly active, oncogenes. But CNA inherently capture background noise: these alterations rarely target a single gene, but may involve large sections of a chromosome. The most successful algorithm to find significantly altered genes from

CNA is GISTIC, which scores alterations by their recurrence as well as their narrowness(Mermel et al. 2011). More localized genetic lesions can be found using whole exome sequencing, which can identify coding mutations that can activate or disable a gene. But much like copy number data, nucleotide sequence data suffers from many passenger mutations. Studies showed that recurrent mutations in a gene could be due to processes unrelated to its role as a driver, including gene length, sequence content, or low constitutive expression (Lawrence et al. 2013). Using recurrence alone to find driver genes has had much success, but many limitations exist(Lawrence et al. 2014).

Besides the technical challenge imposed by the prevalence of passenger mutations, another problem is more basic to the premise that driver genes will be highly recurrently altered across

patients. It is well known that many routes to cancer exist, and this can clearly influence the prevalence of a driver gene across samples. In glioblastoma, tumors have often been classified as primary, occurring in older patients, or a less common secondary type, occurring in younger patients as a progression from a lower grade neoplasm. The two tumor varieties may have no visible histological distinction, but they have been shown to carry different genetic lesions and different prognoses. Using gene expression profiles of samples from TCGA glioblastoma project,

Verhaak et al. clustered glioblastomas into four subtypes, connecting the subtypes to different sets of copy number or point mutations, as well as distinctions in disease phenotype (Verhaak et al. 2010). For example, the proneural group contains more secondary glioblastomas, has different treatment response, and may show better prognosis. Subsequent work further characterized glioblastomas by their methylation profiles, finding a subset of mostly proneural glioblastomas with a distinct profile of CpG island hypermethylation (G-CIMP subtype)(Noushmehr et al. 2010;

Brennan et al. 2013). These had a distinctive clinical phenotype and profile of copy number alterations and mutations, as compared to other proneural tumors, as is shown in Figure 1-1. The authors suggest that these genetic alterations cooperate specifically with the methylation-induced gene silencing in these tumors.

Figure 1-1: Subtypes of glioblastoma.

As clustered by methylation profiles (“DM CLUSTER”), clusters show differences in clinical indicators as well as gene expression clusters (“EXP CLUSTER”) and presence of somatic copy number aberrations or mutations. Reproduced from (Brennan et al. 2013) figure 5.

This type of analysis shows that context is important in understanding driver mutations, and less frequent subtypes, such as G-CIMP tumors, may have their own sets of highly relevant drivers.

Melanoma provides other examples of the heterogeneity among tumors. A variety of alterations occur to activate the pro-growth MAPK pathway: BRAF gain of function mutations occur in over

50% of patients, with NRAS mutations in another 20% or more (Watson et al. 2013). But these two members of the MAPK pro-growth pathway almost never co-occur, either because of lack of selective advantage to further disruption of the MAPK pathway, or because such co-mutation proves deleterious. This type of pattern has led to the “exclusivity hypothesis”, which states that redundant mutations are less likely to be found in the same tumor. Thus, mutually exclusive mutations could be informative of functional relationships between genes. Additionally, the mutual exclusivity and prevalence of these two alterations suggests that activation of the MAPK pathway may be crucial for melanoma development. Intriguingly, some melanoma subtypes harbor no BRAF or NRAS mutations, including acral and mucosal melanomas that often bear activating KIT mutations that may influence the same pathway. It is important to note that no two 5

changes are truly redundant: even NRAS and BRAF, which are adjacent in the signaling network, have different functions. While both NRAS and BRAF mutations activate MAPK, NRAS additionally activates the PI3K pathway(Palmieri et al. 2009). In conclusion, much like in glioblastoma, melanoma shows a variety of pathways to tumor development, and shows how cancer alterations can be informative of each other.

To summarize current work in finding relevant alterations in cancer genomes, much emphasis has been placed on finding the most recurrent changes. But simultaneously, it is understood that many pathways to cancer exist, including subtypes of tumors. Mutations that are less prevalent across the population can still be highly relevant for an individual tumor. Understanding subtypes of cancer is as important a question as finding driver genes, and the two goals are highly interlinked.

Finding subtypes of cancer can help understand pathways to the disease: the subtypes of glioblastoma have been linked to different neural cell types. Understanding the commonalities and distinctions among sets of tumors will provide a more complete picture of tumor biology.

My goal in this dissertation is to find important genes genetically altered in cancer, but also to understand how these lead to tumor development. I apply current tools to find the recurrent genes of interest across compilations of tumor samples, but my main focus is in developing new approaches to using large cancer genomics compendia to understand cancer biology. In chapter 2

I describe work characterizing genetic alterations driving melanoma, the most lethal skin cancer, and a cancer that is incurable in its metastatic form. Due to high rates of ultraviolet induced DNA damage, melanoma genomic profiles are highly complex, with hundreds of protein changing mutations per patient. In newly sequenced cases of melanoma, I work on identifying genes with evidence of selective alteration in the disease. As nevi are a risk factor for melanoma, and

dysplastic nevi are considered to be possible melanoma precursors, I also compare dysplastic and other nevi to melanoma to better understand the progression from a benign nevus to a melanoma.

Beyond current methods for finding recurrent genes driving oncogenesis, I develop novel approaches to identify important genes. I propose in this work that in genomic and clinical profiles of populations of cancer patients contain patterns that are an underutilized source of knowledge about cancer biology. The approaches I develop mine these patterns for new evidence of genetic alterations driving cancer development, with a particular focus on melanoma and glioblastoma. In section 3, I develop methods using measures from information theory to find mutation patterns. In 3.1 I describe an algorithm for finding Genetic Alteration Modules with

Total Correlation, or GAMToC. This method addresses the combinatorial nature of genetic alterations in cancer. The examples above provide ample context for the concept: a certain combination of genetic lesions are present in subtypes of glioblastoma, and MAPK activating mutations are mutually exclusive across cases of melanomas. GAMToC is an information theoretic approach to find these patterns across compilations of cancers, and to exploit these types of patterns to find driver genes and to understand their role in cancer development. I show results of the method in glioblastoma, and, motivated by these observations, I extend this method in 3.2, which describes GAMToC-L. While the first version of GAMToC searched for combinations of highly recurrent genetic alterations across tumor compendia, GAMToC-L uses all genomic information. I consider that a non-random pattern of joint genetic mutation that includes a gene can help us discover new less recurrent drivers in caner.

My work with GAMToC developed an unsupervised method that relies only on patterns of mutation within a collection of cancer samples. In the final section I will bring together multiple sources of information to find driver genes. In this chapter, I discuss the new possibility of using

Electronic Health Records (EHR) to find novel cancer-related genes. Included in supplementary chapter 5 is a brief discussion of the utility of EHR for finding patterns of disease. While

GAMToC looked for combinations of genetic alterations, the approach described in chapter 4 uses combinations of co-occurring Mendelian disease and cancer. As the mutations that are responsible for the Mendelian disease are usually known, I present clinical co-occurrence as a novel source for identifying cancer drivers.

Genomic profiles of tumors, including genetic mutations, gene expression, and epigenetics, can exquisitely characterize a particular tumor in a particular patient. Francis Collins, director of the

National Institutes of Health, has called such high dimensional genomic data “the leading edge of precision medicine.” But precision medicine requires advances in basic science that can identify the pathways that are most relevant in cancer development, and thus the mutations and other cellular changes that likely drive a particular patient’s tumor. Novel approaches to identifying patterns in large datasets will indicate the selective processes shaping cancer, and common ways that tumors overcome these obstacles. Using large patient cohorts, we can thus better understand how the disease arises in each patient, and we can identify vulnerabilities that can be exploited in targeted therapies. This thesis illustrates some of the challenges, solutions, and opportunities created by the technological revolutions in data acquisition and high throughput genomic technology.

2 Coding mutations influencing development of melanoma and nevi

Metastatic melanoma is a highly lethal disease with no effective current treatment. Risk factors include chronic exposure to mutagenic ultraviolet light and the presence of dysplastic nevi, suggesting a progressive evolutionary process leading to the cancer. However, melanoma can also arise from sites lacking sun exposure, and some distinctions in mutation profiles have been associated with both body location and sun exposure. Despite the highly recurrent presence of mutations in BRAF and NRAS, and the understanding of many pathways involved in the disease, the high background mutation rate in melanoma makes identification of driver genes difficult. It is a heterogenous tumor with a highly complex landscape of genomic alterations. Here, we report two approaches to discovery of driver genes in melanoma. First, we perform whole exome sequencing on a cohort of melanomas. We examine the resulting mutation profiles for recurrent alterations, and then we further investigate for genes with evidence of potential as a therapeutic target. The results from this section are published in (Aydin et al. 2014). Second, we sequence a set of nevi, including dysplastic nevi and congenital melanocytic nevi, to find genes that drive nevus development, and genes that distinguish a nevus from a melanoma.

2.1 Sequencing melanomas and discovery of FBXW7 as a melanoma

tumor suppressor

In order to find novel driver genes in melanoma, we perform whole exome sequencing of a small exploration set of metastatic melanomas. We identify a gene, FBXW7, with evidence for a role as a novel tumor suppressor in melanoma. FBXW7 has known interaction with the oncoprotein

NOTCH1. Our results from functional validation and in vivo studies suggest that inactivating mutations in FBXW7 have relevance in indicating notch inhibitors as a treatment for melanoma.

2.1.1 Methods

We screened a cohort of eight metastatic melanomas using whole exome sequencing. Sequencing resulted in an average of 42 million reads per sample (32 to 101 million), of which an average of

98.4% mapped to the hg19 genome using BWA (H. Li and Durbin 2009), followed by GATK indel realignment, resulting in an average depth of 11 reads per base covered at depth greater than zero. Using the SAVI algorithm (Trifonov et al. 2013), we called positions with nucleotide mutations. From these, we retained only the variants at positions with depth greater than 10 in both tumor and normal samples, and we filtered out any variants that also appeared in any normal sample in a greater than 25% of reads. We identified of a total of 2308 exonic mutations, with

737 synonymous and 1571 non-synonymous, consisting of 1431 missense and 78 non-sense mutations, and 62 insertions/deletions. The mean exonic non-synonymous mutation rate was

10.6 mutations per megabase, with mutation rates varying from 2.8 to 26.7. All cases sequenced were cutaneous melanomas on sun-exposed sites, and as expected the majority of nucleotide substitutions were C>T or G>A transitions (73-91% of all mutations), indicative of ultraviolet- induced damage as well as cytosine deamination. The hot spot mutation, BRAF V600E, was present in six of the eight cases.

Following sequencing and variant calling, we used the collection of mutations to identify genes with evidence of positive selection for nonsynonymous mutations. First, we evaluated whether a gene had more nonsynonymous mutations than would be expected. We estimated the expected number of nonsynonymous mutations for each gene using the number of synonymous mutations

in that gene, NS,G and the nonsynonymous to synonymous mutation ratio across all genes, NN/NS resulting in an expected number of nonsynonymous mutations of NS,G*NN/NS. Then we evaluated whether the observed number of nonsynonymous mutations, NN,G was significantly more than this expected value using a Poisson model. We also tested for elevated number of mutations given

gene length, using the background number of mutations per coding positions. For this test, we use a binomial model, with amino acid length corresponding to number of trials, and probability of a nonsynonymous mutation calculated from the average number of nonsynonymous mutations per amino acid, across all genes with mutations. Finally, our candidate list contains 23 putative driver genes that have p < .05 in both tests (Supplementary Table 1).

As we were particularly interested in finding therapeutically related mutations in melanoma, we searched for genes that might be impacted by our mutations, and that are druggable targets. To find interacting partners for the mutated genes, we use GeneRIF interactions(Brown et al. 2005), keeping only those interactions that are documented in human cells and are not based on affinity assays. Then, for each of these interacting partners, we use the Cancer Commons(Shrager,

Tenenbaum, and Travers 2010) drug target database to assess if partners are druggable. Only a few recurrently mutated genes, including FBXW7 have druggable interacting partners.

2.1.2 Results Using exome sequencing we first call variants and then select the genes with evidence of positive selection for their alteration. Then, we use external sources of evidence about gene interacting partners to identify FBXW7 as a putative melanoma driver of interest. We further support a role for FBXW7 by sequencing a wider panel of 103 melanomas including 77 tumor samples and 26 cell lines. We sequence the coding regions of FBXW7, BRAF, and NRAS. Non-synonymous mutations in FBXW7 appear in eight cases (8% frequency), with five nonsense, two missense, and one frameshift mutation. This is a significantly elevated mutation rate: the probability of having this number of nonsynonymous mutations is less than 10-4, given the length of the gene (710 amino acids) and the nonsynonymous mutation rate per base in our samples (1 x 10-5). Mutations within the WD40 domain of FBXW7 are predicted to disrupt substrate binding, and thus lead to sustained activation of its substrate oncoproteins. Of note, the presence of mutations in FBXW7

does not correlate with BRAF or NRAS mutation status. Collectively, these findings identify somatic mutations of FBXW7 as a novel recurrent genetic event in melanoma.

In work to characterize FBXW7’s disruption in the disease, we profile expression of the gene in a panel of melanomas as compared to benign nevi, and we find that FBXW7 is downregulated in melanoma, and underexpression correlates with mutation. As FBXW7 has a demonstrated role in other tumor types, particularly in its interaction with known oncoproteins(Oberg et al. 2001;

Welcker et al. 2004), this gene is of high interest.

Then, to investigate the mechanism of FBXW7’s influence on melanoma, we examine the effect of FBXW7 loss on known regulated proteins including NOTCH1, its direct target, and other targets including CCNE1 and MYC. Of these, NOTCH1 is consistently upregulated in cell lines with loss of FBXW7. As well, we transplanted immunodeficient mice with NRAS mutant melanocytes, and then used an shRNA targeting FBXW7 to silence its expression. In this xenograft experimental model, not only is NOTCH1 significantly upregulated, but tumors grow at accelerated rates compared to a control shRNA (Figure 2-1). Ectopic expression of the mutant

FBXW7 in the NRAS mutant melanocytes also accelerates tumorigenesis. As NOTCH1 appears to be strongly regulated by FBXW7, and this change appears to influence tumor growth, we create a set of xenografts bearing the NRAS mutation and FBXW7 knockdown, and we treat the resulting tumors with a notch inhibitor, dibenzazipine. These tumors show significant reduction in growth as compared to a control group. Thus, this study suggests a mechanism for activation of

NOTCH1 in melanoma, via the newly identified melanoma tumor suppressor FBXW7, and suggests that in some melanomas with deregulated NOTCH1 expression, as via FBXW7 ablation, notch inhibitors may be a useful therapy.

Figure 2-1: FBXW7 as a novel therapeutic target in melanoma

These figures are reproduced from (Aydin et al. 2014). A. mutants of FBXW7 are associated with larger tumor volume. B. mutants of FBXW7 are associated with greater expression of its ubiquitinated target NOTCH1, and with upregulation of NOTCH1’s target HEY1. C. Treatment with Dibenzazipine (DBZ) significantly reduces tumor volume. 2.2 Sequencing nevi: exploring the progression to melanoma

To further explore factors influencing development of melanoma, we explore the genomic landscape of nevi and dysplastic nevi. Dysplastic nevi are benign neoplasms of melanocytes that are considered both risk factors for and possible precursors to melanoma. Patients with a dysplastic nevus have twofold risk for melanoma, while patients who bear more than ten dysplastic nevi have a 12-fold increased risk(Elder 2010). Although many melanomas arise de novo, about 25-50% of melanomas have a histologically associated nevus. As melanoma is thought to result from progressive alterations, we wished to characterize the genomic landscape of nevi. The goal is to identify what genetic events separate a benign from malignant state, in related tissue types. Our panel includes multiple nevi per patient for some patients, as well as matched normal blood samples for each patient. In this project, we confirm the widespread presence of the known nevus driver mutations affecting NRAS and BRAF. We also show that nevi from the same patient display a branched pattern of evolution. Finally, we identify genetic changes that are distinctly significantly less likely to occur in nevi as opposed to melanoma, pointing out sets of events the precipitate malignant transformation.

2.2.1 Methods 2.2.1.1 Calling significant variants in nevi

We collect 35 nevi from 22 patients, including eight common acquired nevi (CAN), four congenital melanocytic nevi (CMN), and 23 dysplastic nevi (DNS) from patients with dysplastic nevus syndrome. We tailor a variant calling strategy to the unique nature of this data: obtaining pure sample of melanocytes from a nevus is impossible, and purity varies between nevi. Due to the predicted impurity of the samples, we first examine all nevi for the presence of BRAF V600 and NRAS Q61 mutations. These mutations are present in a variable fraction of reads aligning to their respective genes, as low as one read, but in multiple reads for 32 of 35 nevi. Thus, we employ a strategy suited to identifying mutations supported by a low percentage of reads. First, conservative alignment to the reference genome is used. BWA (H. Li and Durbin 2009) with no

Smith-Waterman mate rescue is followed by realignment of reads that may have been misaligned due to insertions or deletions. Next, presence of variants is called using the SAVI statistical procedure (Trifonov et al. 2013). Variants are filtered by the SAVI statistic, strict absence in the matched blood sample, as well as using read depth, and the presence of supporting reads aligning to both strands. Another filter uses a sample-specific threshold on the frequency of variant-calling reads among the reads aligned to the variant position. This sample-specific threshold is calculated by using the BRAF V600E (or NRAS Q61K/L) variant read depth in that sample as approximation of the expected heterozygous variant presence per sample. From this presence a lower bound on expected heterozygous variant frequency is determined using a binomial to model the read distribution. A minimum frequency of 3% variant is imposed. Finally, presence of each variant is checked using SamTools (http://samtools.sourceforge.net/SAM1.pdf) quality greater than one, which takes into account other characteristics of reads that identify a position as a true variant.

2.2.1.2 Comparison of nevus mutation to melanoma mutation

We download level 2 somatic mutations from TCGA, as well as the results of the Broad Institute

MutSigCV pipeline. We use MutSigCV’s statistic(Lawrence et al. 2013) to identify those genes with a significantly recurrent mutation pattern in melanoma. Then, we test whether each gene is significantly more frequently mutated in the melanoma cohort as opposed to the nevus cohort.

This can be tested using a binomial model of nonsynonymous mutation frequency per nevus or melanoma. The difference between the binomials could be calculated as a chi-square statistic, or similarly, using the hypergeometric statistic to test whether nevi are significantly depleted for a given mutation. This quantifies whether the melanoma genes have a similar, or lesser, rate of mutation in nevi, as compared to the gene’s mutation rate in melanoma.

2.2.2 Results and discussion 2.2.2.1 Spectrum of mutations in nevi We collect 35 nevi from 22 patients, including eight common acquired nevi (CAN), four congenital melanocytic nevi (CMN), and 23 dysplastic nevi (DNS) from patients with dysplastic nevus syndrome. As described in the Method, we combine a liberal sample-specific threshold for identifying mutations with stringent quality filters that remove many of the variants most likely to be false positives.

There are a median of 16 nonsynonymous and 9.5 synonymous mutations per nevus, but mutation profile varies widely, ranging from four nevi with no nonsynonymous mutations to a nevus with

61 nonsynonymous mutations. Broken down by subtypes, two of the four CMN have no mutations. The other two have five and four mutations, making this type of nevus the least mutated. Both CAN and DNS have higher mutation rates. CAN have a median of 10 nonsynonymous mutations, with all cases displaying one or more mutations. Mutations rates in

DNS are higher, as might be expected, with a median of 21 mutations per case. However, two of the DNS nevi have no called mutations, possibly due to low purity. For the dysplastic nevi, CàT mutations are predominant, consistent with UV induced damage and cytosine demination 15

mechanisms. Common acquired nevi largely display a similar pattern, while mutations in congenital nevi, although rare, do not appear to share this pattern (Figure 2-2).

Numbermutationsof Fractionmutationsof DNS CAN CMN 30 20 10 Threshold 0

Figure 2-2: Mutation spectrum of nevus types, and frequency thresholds derived per nevus

To identify potential novel genes beyond BRAF and NRAS that could contribute to nevi, we compile a list of genes with recurrent alterations. 22 genes are found mutated in two or more nevi.

The list includes long genes known to be commonly somatically altered (e.g. SYNE1, DNAH5), due to unknown mechanisms. The other genes that present higher mutation rate are: NRAS,

BRAF, NOL4, TEK, PCDHB14 and PCDH15. Interestingly, NOL4 is epigenetically silenced in squamous cell carcinomas of the head and neck, making this a potential tumor suppressor in these cancers(Demokan et al. 2014). TEK is a protein tyrosine kinase that is most associated with endothelial cell growth signaling and vascular development. Both PCDHB14 and PCDH15 are members of the protocadherin family that are most expressed in neural cell junctions.

2.2.2.2 BRAF or NRAS may be mutated in all nevi Because BRAF and NRAS are known to have activating mutations in melanoma as well as in dysplastic and congenital nevi, we first examine the presence of these activating mutations in our collection. We find statistically significant presence of BRAF V600E mutation in 14 nevi, and

NRAS Q61 mutation in four nevi, including two Q61K and two Q61R mutations. These mutations are present a wide range of frequencies, from 17% to 58% of reads covering these regions of the exome. This wide range is to be expected given the impurity of the melanocyte content in the nevus sample, when combined with possible subclonal mutation load and sequencing error.

We examine how BRAF and NRAS mutation are associated with subtype, and whether we can rule out mutation to these loci in the 17 nevi with no BRAF or NRAS mutation called. Thus, we check the reads aligning to the two mutation regions for any presence of these mutations. High quality sequencing reads supporting the relevant mutations are present at very low frequency in all remaining nevi, in greater than 4% of the reads in most cases. For the four CMN, two have

called NRAS Q61K/R mutation and the other two have evidence for mutated NRAS at low frequency. The CAN all have strong BRAF V600E mutation.

The DNS set contains multiple samples of low purity, complicating efforts to ascertain mutation rates. Of the 25 DNS, six have strong BRAF V600E mutation and two have strong NRAS

Q61K/R mutation. Twelve more of the DNS have BRAF variants called at high quality but in a low fraction of reads. One more dysplastic nevus has credible evidence of NRAS Q61K mutation, again at low frequency. The remaining two dysplastic nevi display possible subclonal composition based on BRAF and NRAS loci: patient P025-nevus-1 has reads supporting both

NRAS Q61K and BRAF V600E mutations, though the NRAS reads are at very low frequency.

Finally, patient P016 has a very interesting pattern of mutation: like the others, this nevus has low frequency chr7:140453136 AàT mutations, present in four reads, which would confer V600E mutation. However, two of those reads also have chr7:140453137 CàA mutations, which could reflect a subclone expressing BRAF V600D. As a comparison, we test the frequency of mutations to nearby codons, BRAF G604 and NRAS D65. No mutations are found.

The results support mutual exclusivity between BRAF and NRAS within a nevus. Using the more liberal thresholds for variant calling, we find that only one of the 35 nevi has both BRAF and

NRAS hotspot mutations of high quality and at a greater than 1% frequency (p-value for mutual exclusivity is 4.9x10-7).

2.2.2.3 Evolution and recurrent mutations in nevi In a subset of patients, multiple nevi are sequenced. This includes patient 8, who presented with both a congenital nevus and a dysplastic nevus, and patient 6, a classic case of dysplastic nevus syndrome. We compare nevi from the same patient to each other to investigate two hypotheses about their pattern of evolution. First, we assess whether the nevi could have any common

precursor cell subsequent to their shared germline. Second, we consider whether the shared genetic background of nevi has an influence in the sets of mutated genes that result in this abnormal state.

In order to discover whether nevi share a common post-germline ancestor, we look for identical mutations in the nevi that are absent in the matched blood sample. As identical BRAF and NRAS mutations are frequent among all nevi, these are excluded. Another identical mutation to

ASPHD1 is most likely due to germline variants or sequence alignment problems. Thus, we conclude that different nevi in the same patient evolve in a branched evolutionary pattern.

Next, we look for overlap in the sets of genes mutated in nevi from the same patient, as an indicator of a shared route to nevus development. We find that two CAN from patient P019 both have mutations to different positions of NOL4. As the nevi have 13 and 15 mutations in all, they are highly unlikely to share mutations to this gene by chance. Similarly, two DNS from patient

P6 have mutations to TEK, while the nevi have only 2 and 14 mutations each. These findings support the hypothesis that convergent evolution occurs in nevi, similar to cancer, and additionally put forward these two genes as being of particular interest in the development of nevi.

2.2.2.4 Melanoma-specific genomic alterations As the intention of this study is to better understand the development of melanoma, and the necessary mutations for a nevus to progress to a melanoma, we compare prevalence of the mutations in nevi to that in melanoma. First, we extract a list of genes with evidence of positive selection for nonsynonymous mutation, using a compilation of whole exome sequencing results from 297 melanoma cases from TCGA, and we estimate the frequency of altered cases from the

297 samples. We statistically determine if nevi have a significantly lower rate of mutation to any

genes, identifying genes that could be the drivers in the transformation from the benign precursor nevus to the cancer. The full list is shown in Supplementary Table 2. We find that TP53,

CDKN2A, and NF1 are less mutated in the nevi. A number of other interesting genes are highlighted, including the TP53 related gene TP63, BCLAF1, another apoptosis related gene.

Other genes on the list are related to epithelial cell junction, including DSG3 and COL3A1.

Our results suggest that key changes in the progression of melanoma to nevus include changes related to the apoptosis response pathway (CDKN2A, TP53, TP63, BCLAF1) and the dermal- epidermal junction (DSG3, COL3A1). These results can be applied in a therapeutic setting aimed at understanding if an abnormal tissue sample is a nevus, or a potential melanoma requiring a more aggressive intervention. Future efforts will help us understand what lesions drive the transition from nevus to aggressive melanoma.

2.3 Discussion

Melanoma is both a clinically and genetically complex disease, and in its metastatic form it is incurable. These studies have made steps toward untangling the genomics of this cancer. In a clinical context, a targeted sequencing panel including FBXW7 could eventually influence treatment decisions such as use of a notch inhibitor. For patients with dysplastic nevi, targeted sequencing could ascertain whether a nevus displays any of the mutations that are associated with melanoma, suggesting a benefit from more aggressive early treatment. However, many more questions could arise as a result of our findings. It would be interesting to know whether

FBXW7-mutated patients display any particular clinical phenotype, or if they are less likely to have amplifications of notch genes. This pattern would be expected if FBXW7 inactivation is sufficient for oncogenic activation of notch. It would also be highly interesting, though practically difficult, to sequence a nevus before and after transition to melanoma. We do not know if any of the mutations in nevi, other than the BRAF and NRAS mutations, are propagated in 20

melanomas. While these small studies allow exploration of specific biological phenomena, with larger cohorts we can find more complex and informative patterns in cancer data.

3 Applying the total correlation to identify and contextualize driver alterations

While my work in melanoma ranks important cancer genes by mutational recurrence across compendiums of tumor samples, landscape mutational approaches that score each gene individually ignore the known effects of mutational context on selection and do not address the combinatorial complexity of genomic alterations in tumors. Tumors are the result of accumulated genomic alterations that cooperate synergistically to produce uncontrollable cell growth.

Identifying recurrent alterations among large collections of tumors is only one way to pinpoint genes that endow a selective advantage in oncogenesis and progression. In my dissertation work I have intended to go beyond recurrence to find how combinations of genetic changes influence the development of cancer. In this section, I develop an information theoretic framework that integrates copy number and mutation data to identify gene modules with any non-random pattern of joint alteration. A non-random pattern of co-mutated genes is evidence for selective forces acting on tumor cells that harbor combinations of these genetic alterations. Although existing methods have successfully identified mutually exclusive gene sets, no current method can systematically discover more general genetic relationships with no prior knowledge. I develop a framework and methods termed Genomic Alteration Modules using Total Correlation

(GAMToC), to find combinations of recurrently altered genes with a related pattern of mutation.

Additionally, I present the Seed-GAMToC procedure, which uncovers the mutational context of any putative cancer gene. All software is publicly available. I apply GAMToC to glioblastoma multiforme samples, and the results show distinct subsets of co-occurring mutations, suggesting distinct mutational routes to cancer and providing new insight into mutations associated with

Proneural, Proneural/G-CIMP, and Classical types of the disease. Indeed, considering combinations of genetic mutations in cancer is a powerful approach to learning about the disease.

This work is under review at the Journal of Molecular Cell Biology. Then, I describe, a follow-up work, in preparation for submission. This works uses the same principles as in GAMToC, but instead this approach enables us to find driver mutations in cancer by their pattern of related mutations. Finding driver alterations in copy number data has proved highly challenging, and the results suggest new sources of evidence for selected alterations in cancer.

3.1 An information theoretic method to identify combinations of

genomic alterations that promote glioblastoma

3.1.1 Introduction Tumors are known to evolve by acquiring genetic lesions. Each mutation creates a cellular state uniquely predisposed to thrive with the addition of further specific survival abilities (Hanahan and Weinberg 2011). Recent studies have successfully exploited the selective pressures on developing tumors to rank important cancer genes by mutational recurrence across compendiums of tumor samples (Beroukhim et al. 2007; Mermel et al. 2011; Lawrence et al. 2013). But approaches that score each gene individually ignore the known effects of mutational context on selection. Tumor survival can be promoted by damage to only one of a set of alternate genes in a pathway (mutual exclusivity of aberration), while other genetic changes only provide a selective advantage to a cancer in a given mutational context (co-occurrence of aberration). For example, in melanoma, BRAF gain of function mutations occur in 40% of patients and NRAS mutations in

25%, but these two members of the MAPK pro-growth pathway almost never co-occur, either because of lack of selective advantage to further disruption of the MAPK pathway, or because such co-mutation proves deleterious (Davies et al. 2002). Despite their frequency, MAPK- activating mutations alone are an evolutionary dead end for the cancer, resulting in cell senescence(Michaloglou et al. 2005). Cancer progression also requires disruption of a tumor suppressor function such as CDKN2A(Michaloglou et al. 2005). This example shows that complex patterns of mutual exclusivity and co-occurrence of mutation, thus far identified in a

piecemeal fashion, are to be expected across cancer cases. Additionally, the observed mutational relationships of genes, and thus the context in which a genetic aberration is of benefit to tumor development, can provide insight into the functions of genes that are altered in cancer.

However, most approaches seeking relationships between cancer mutation events focus on mutually exclusive lesions, reasoning that this pattern may reflect underlying pathways(Mark D.

M. Leiserson et al. 2013; Vandin, Upfal, and Raphael 2012; Szczurek and Beerenwinkel 2014; C.

A. Miller et al. 2011). But these methods will miss other relationships between mutations, such as co-occurrence. Additionally, the assumption that different genes in the same pathway are interchangeable is a strong claim. Combinations of genes have been found to jointly predict cancer phenotype (Varadan and Anastassiou 2006; Mo et al. 2013), but, to our knowledge, no unsupervised method exists for finding related genetic alterations.

A different approach has been to scan for representation of dysregulated genes within gene sets known to be functionally related. Recent studies have found pathways predicted to be perturbed by differential gene expression (Tarca et al. 2009), or mutation (Boca et al. 2010), or when multiple sources of information on gene activity are integrated (Vaske et al. 2010). Other methods have used graph topology to find functional interaction sub-networks enriched in mutated genes(Vandin, Upfal, and Raphael 2011; Cerami et al. 2010; Hofree et al. 2013; G. Wu,

Feng, and Stein 2010), or to identify cliques of genes with mutually exclusive mutational occurrence(Ciriello et al. 2011). These approaches have the advantage of being able to use diverse genome-wide alteration information and provide a biological context for the patterns discovered, but they rely on known gene interactions and on narrow definitions of gene interaction.

We propose a method that integrates copy number and point mutation information, does not require prior functional information, and can find any structured module of genes, rather than only mutually exclusive alterations. The method, Genomic Alteration Modules using Total

Correlation (GAMToC), selects a gene set with high total correlation. Total correlation measures the difference between the joint uncertainty, or entropy, of a set of variables (genes), as compared to their individual uncertainties(Watanabe 1960). When there is no joint relationship between the variables, the difference will vanish. On the other hand, a high total correlation suggests a joint relationship among the variables, which is not necessarily linear. Because our method can detect any sort of dependency between the variables, it is sensitive to unexpected varieties of gene interactions. It does not require the assumption that different alterations to the same pathway are more or less interchangeable, and it is not restricted to finding genes only in the same pathway.

Instead, the genomic data can lead us to the combination of functional changes that are cooperating in the cancer. We present two implementations of GAMToC, one that uses a greedy method to find a single module starting from a pair of related genes, and another that uses a

Simulated Annealing (SA) method to find the highest-scoring gene set. We examine the speed of the two implementations as compared to exhaustive search, and we evaluate their sensitivity in simulated data. Then, we apply the method to glioblastoma multiforme (GBM) copy number and mutation data from the TCGA. Additionally, in Seed-GAMToC we make use of the same principles to characterize query genes with a likely, but unclear, role in cancer progression by finding a module that contains genes with a related pattern of selection.

We apply GAMToC to copy number and nucleotide mutation measurements from The Cancer

Genome Atlas (TCGA) glioblastoma project (“Comprehensive Genomic Characterization Defines

Human Glioblastoma Genes and Core Pathways.” 2008), as summarized in Figure 3-1. We are able to recapitulate known gene interactions, and we additionally recover genes associated with

subtypes of glioblastoma. Our results suggest that specific alterations to key cancer pathways are not equivalent: on the contrary, there are clear contexts where functionally related genes are differentially selected for alteration. Thus our method is uniquely suited to find and characterize genes that are related in cancer development. The software is freely downloadable and can be applied to any copy number and point mutation data set.

Tumors Binary matrix Somatic Mutation describing patients and mutated genes Somatic Copy Number t1 t2 t3 …

Genes Alteration g1 g2 g3 g4 …

t1 t2 t3 … t1 t2 t3 … Greedy method, g1 t1 t2 t3 … g1 g1g2 g1# n simulated annealing, g2 g2# g2g3 g3 H(gi ) and visualization of g3g4 ∑ g4 g3# i=1 GAMToc Module as g4… … g4# pairwise network …

t1 t2 t3 … g1# g1 g2# g2 H(g1, g2,!, gn ) g3 g3# g4 g4# … n TC(g1, g1,!, gn ) = ∑H(gi )− H(g1, g2,!, gn ) i=1

Figure 3-1: Workflow of GAMToC gene set finding.

Genomic alterations (e.g., CNAs and somatic mutations) are integrated to create a binary matrix of samples and genes. The total correlation score compares the entropy of the mutational statuses of individual genes (labeled 1 through 4) against their joint entropy, in effect testing the hypothesis that these gene mutational statuses have a relationship (indicated by the connected network). GAMToC finds sets of mutationally related genes using this score, and we visualize the results in a pairwise correlation network.

3.1.2 Method 3.1.2.1 Preprocessing genetic aberration data Currently the GAMToC algorithm can start from assessments of sample copy number aberrations and from nucleotide variant calls resulting from whole exome sequencing (WES) data. For the

TCGA GBM data, we downloaded processed data from the Broad Institute Firehose

(http://www.broadinstitute.org/cancer/cga/Firehose) download data set of 9/23/2013. This includes mutation calls, GISTIC2 results, and thresholded calls of copy number status per gene per tumor. Both copy number and matching WES data was available for 273 GBM patients.

For copy number data, we remove calls in regions of copy number polymorphism, as called by the Broad Institute pipelines, and we keep only copy number alterations in genes that are in called

GISTIC2 peaks. For the nucleotide variant calls, we record any gene with a somatic nonsynonymous mutation as mutated in the patient. The result of this initial step is a binary matrix of patients and genes that marks patients as having a mutation in a gene.

We combine the two matrices in an "or" gate fashion. Finally, we merge genes on the same chromosome that are altered in exactly the same samples into a single unit. It is important to note that copy number aberrations are usually not focal events targeting a single driver gene, and in fact often involve entire chromosomes. Thus, even distant genes on the same chromosome as another gene already included in the module will score as the best candidates for module inclusion, although this does not reflect any functionally interesting genetic interaction (Figure

3-2). In order to remove this bias, we do not allow any module to contain more than one gene from the same chromosome.

PRDM2& MDM4& Column1 no(PRDM2 PRDM2 no(MDM4 187 10 MDM4 47 29

P&=&4x10,12&&

Figure 3-2 Illustration of copy number linkages in the GBM cohort

The entire chromosome 1 is shown. Two significantly recurrently amplified genes are shown, MDM4 and PRDM2. While both events are selected for amplification, and chromosome 1 is the largest chromosome, it is impossible to distinguish significantly co-occurring events from the effect of the amplification of the entire chromosome. 3.1.2.2 Scoring the module Our aim is to find the most mutually informative set of genes, using the total correlation score:

�� !, �!, ⋯ , �! = �(�!) − � �!, �!, ⋯ , �! !!!

To find the significance of this value, we apply the G-test as follows.

The total correlation, or mutual information, of two variables x1 and x2 can be reorganized to form the Kullback-Leibler divergence from the joint distribution, p(x1 ,x2) of the independent distribution, p(x1 )p(x2). As outlined in (Goebel et al.), we can treat the Kullback-Leilbler divergence (and therefore mutual information) as a function of the joint p(x1 ,x2), and expand this as a Taylor series about the point p(x1 ,x2) = p(x1 )p(x2). The resulting expression, using the expansion terms only up to order 2, is identical to that of a chi-squared statistic, when multiplied by N (the number of data points, to convert probabilities to count-equivalents), and 2ln(2), accounting for the change of base from base 2, and the coefficient of the Taylor series expansion. 28

The degrees of freedom of this chi-squared statistic would be 1, in our case of mutual information of two binary variables. To give an example of the calculation of the degrees of freedom, if we have two genes in the module, there are four possibilities, which can be viewed as a two by two contingency table as below:

x2 non-mutated x2 mutated x1 non-mutated x1 mutated

This table has 22 = 4 cells with 3 constraints, consisting of the number of mutations per each of two genes, and the total number of samples. Thus, there is one degree of freedom.

The multi-variable total correlation is an extension of the deduction for mutual information.

Using the same logic as outlined above, we can reformulate the total correlation formula as

n TC(X1, X2,, Xn ) = ∑H (Xi ) − H (X1, X2,, Xn ) i=1 1 p x , x ,, x p x , x , , x ln X1X2Xn ( 1 2 n ) =  X1X2Xn ( 1 2  n ) ln2 ∑∑ ∑ p (x )⋅ p (x )⋅⋅ p (x ) x1 x2 xn X1 1 X2 2 Xn n 2 $p x , x ,, x − p (x )⋅ p (x )⋅⋅ p (x )& 1 % X1X2Xn ( 1 2 n ) X1 1 X2 2 Xn n ' 3 =  +O 2ln2 ∑∑ ∑ p (x )⋅ p (x )⋅ ⋅ p (x ) x1 x2 xn X1 1 X2 2  Xn n 2 $n x , x , , x n(x ) n(x ) n(x ) / N n−1& 1 % ( 1 2  n ) − 1 ⋅ 2 ⋅⋅ n ' 3 =  +O 2N ln2 ∑∑ ∑ n(x )⋅ n(x )⋅ ⋅ n(x ) / N n−1 x1 x2 xn 1 2  n

3 n x , x ,, x where O is the Taylor series remainder term of order 3; ( 1 2 n ) is the observed number

n(x )⋅ n(x )⋅⋅ n(x ) / N n−1 of events, and 1 2 n is the expected number of events. Again, according to the chi-square test, 2N ln2 ⋅TC approximately follows a chi-squared distribution, with degree

n of freedom, for n binary genes mutation statuses, of 2 − n −1 (Kullback 2012) (correct only

when the number of samples is bigger than 2n). As in our example of two binary genes above, the formula calculates the degrees of freedom as 22 - 2 – 1 = 1.

Actually, total correlation is a special case of the G-test. In statistics, G-tests are formulated as

�! � = 2 �! ∙ ln ( ) �! !

O E where i is the observed distribution (frequency), and i is the expected distribution based on null assumption. It can be proved that G approximately follows a chi-squared distribution (Sokal and Rohif 1981).

It is important to mention that the number of samples is important to the approximation of the distribution of total correlation. We simulate five independent variables with different number of samples ranging from 2 to 100. The theoretical value approaches simulation results very well when the number of samples is larger than 20, but the G-test fails when sample size is small.

Therefore in our application of our total correlation method, if the number of samples is larger

n than the degrees of freedom, 2 − n −1 , we can use the G-test. Otherwise, we must use a permutation method to calculate the p-values.

3.1.2.3 Module selection The greedy method starts from the pair of genes with the highest mutual information. To grow the module from this initial pair of genes, we then test each other remaining gene to find one, which, together with the existing gene set, will create a set with the highest total correlation. If no module is found at a greater significance level than .05 divided by the number of genes remaining in the module, growth is terminated. We continue to add genes until reaching the maximum feasible module size where joint entropy can be estimated, which is less than the logarithm of the number of samples.

The goal of the SA method is to sample modules of genes in proportion to the total correlation of the modules. The GAMToC SA starts from any initial gene set of a selected size. We use the maximum feasible module size for G-test calculations, given our sample size. For the GBM combined copy number and whole exome data set of 273 tumors, this is a module of eight genes.

The chain continues at each iteration by randomly choosing a gene from the module and replacing it with another gene chosen at random from the non-module genes. If the score of the module is improved by this replacement, then the replacement is retained. If instead the new gene creates a decreased total correlation, the module change has a probability of being retained

(paccept), according to the change of the total correlation. We define log (paccept) as proportional to the change of total correlation, with a proportionality constant that we define as 1/temp.

The temperature starts as "hot", such that a small decrease in total correlation results in a likely probability of acceptance. The temperature continues to decrease by a percentage after a minimum number of iterations and a minimum number of changes to the module. After the change is retained or discarded, the resulting module is the next state in the chain. If the annealing process stops for a certain number of iterations, it will restart at the highest total correlation module that was reached in the course of the annealing, and continue the process at the current temperature. The process will continue until the annealing converges. The final highest total correlation module is our solution.

1 0.1 0.8 0.05 0.6 0.01 0.005 0.4 0.001 p − accept 0.2

0 0 0.1 0.2 0.3 0.4 delta−TC Figure 3-3: Visualization of the relationship between temperature (in legend), change in total correlation (x-axis), and acceptance probability in the simulated annealing (y-axis). 3.1.2.4 Simulation of module and assessment of results For the simulation, we chose to create a data set of 100 genes and 100 patients, and we embed a six gene module in this data. Thus, each simulation creates a binary matrix of gene mutations per patient. For the embedded module, the simulation uses a parameter specifying the fraction C of the patients that are covered in the module pattern, where the rest of the patients have no module pattern. The other parameter specifies random noise N added to the module genes.

First, we simulate the background mutations for independently mutated genes. On average in the glioblastoma data each gene is mutated in 12.9 samples, with a steep decline in number of genes with higher mutation rates. We model this distribution with an exponential, with the empirical mean value as the distribution parameter. Sampling from this exponential distribution, we simulate the background independent mutation rates for each gene. Then we generate the mutations for each patient for each gene as a Bernoulli process according to that gene's simulated mutation rate. Next, we embed in this data a module covering C patients. We generate an exclusive or triplet for the first three genes. We use a multinomial distribution, based on the mutation frequencies of the three genes, to pick which two of the three would be mutated for each covered patient. The final three genes are the negation of the first three genes. Then, according to the noise, N% of the module bits are flipped. 32

For each simulation, the greedy module and the simulated annealing module are assessed, and we compare how many of the 6 genes are recovered in each of the 100 simulations for each parameter setting (Figure 3-5C).

3.1.2.5 Comparison to tumor classifications After obtaining our results, we compare the genes included in our module to previous classifications of tumors. This is motivated by the clear correspondence of our module to aspects of these previous tumor classifications, such as the association of Classical tumors with EGFR and the G-CIMP with IDH1. Tumor classification performed by the TCGA in (Brennan et al.

2013) was downloaded from http://tcga- data.nci.nih.gov/docs/publications/gbm_2013/supplement/Molecular_subtype_classification.xlsx.

Of the patients included in our study, 233 were classified in that work. We compared these classifications with mutation status of each module gene, in order to assess whether the mutations were markers of GBM subtypes.

3.1.3 Results 3.1.3.1 Utility of searching for mutually informative gene sets While many well-characterized cancer driver genes are highly recurrent, more rarely mutated tumor drivers are difficult to identify amidst unstable genomes when using mutational frequency alone. Thus, we must utilize other aspects of the alteration pattern of these genes, such as mutual exclusivity or co-mutation with other genes, keeping in mind that frequency of individual lesions may be low.

As can be seen in Figure 3-4A the number of samples needed to statistically identify mutual exclusivity between a pair of genes grows large when the frequency of mutation is low, and this size is orders of magnitude larger than the number needed to identify co-mutated pairs. This is intuitive as the expectation is that two infrequent mutations are most likely to have no co-

occurrence. When a set of mutually exclusive genes, each with the same low mutational frequency, is instead assessed for a significantly related mutation pattern, the number of samples required to attain significance is much lower (Figure 3-4B).

A Mutually exclusive mutations

Co-occurring mutations

Mutually exclusive mutation set

C Total correlation p-value = .004

genes samples Pairwise correlation p-value = .76

Figure 3-4: Ability to find multi-gene co-mutational patterns.

A. Finding mutually exclusive pairs of gene mutations requires orders of magnitude more samples as compared with finding co-mutated genes. B. With a larger set of genes, fewer samples are needed. C. For an exclusive or triplet pattern, the total correlation is strong, but a pairwise correlation or anti-correlation score would fail to detect a relationship.

Additionally, multi-gene patterns may exist other than mutual exclusivity or co-occurrence. An example would be an “exclusive or” triplet of genes where lesion of any two of the genes is enough to change a phenotype, and the third adds no further advantage. As is shown in Figure

3-4C, the total correlation of this three-gene pattern is highly significant, but the genes display no mutual exclusivity or co-occurrence pattern.

3.1.3.2 Evaluation of greedy and simulated annealing algorithms We have implemented two methods that integrate copy number and point mutation data to find sets of genes with high total correlation, both taking different approaches to finding patterns in this data. The greedy method finds a module by starting from the pair of genes with the strongest mutual information, iteratively adding the gene that creates the best score. On the other hand, the

SA method allows us to explore the broader landscape of modules in order to find an optimal solution. In general, SA methods semi-randomly sample possible solutions to a hard problem, sampling those with the better scores (objective function) more often. Our application of SA samples combinations of genes with high total correlation, and it can find a solution with a higher score. A detailed description can be found in the Methods section.

First, we compared the running time of our implementations against each other and against an exhaustive method. We create a simulated data set containing 100 genes and 100 samples. As shown in Figure 3-5A, time complexity of the exhaustive method increases exponentially with module size, while the greedy method will finish in tens of seconds and the SA method will finish in tens of minutes.

To evaluate the accuracy of the greedy and SA approximations, we randomly generate an embedded module in randomly simulated data, as described in 3.1.2.4. This simulated module has a 6 gene pattern including an exclusive or triplet of genes and their negations (Figure 3-5B), while all other genes are randomly mutated at an exponentially distributed background mutation rate. Two simulation parameters are used: coverage and noise. In a larger coverage, most patients contain this pattern for the module genes, while the rest of the

A B 1000000 Simulated Annealing 1000000 Greedy 100000 Exhaustive g6 100000 10000 g5 100010000 g4 seconds 1001000 g3 seconds 10 g2 100 g1 10 3 4 5 6 module size

3 4 5 6 module size C 6

4 20% covg,greedy 20% covg,SA 50% covg,greedy 3 50% covg,SA 80% covg,greedy 80% covg,SA 2

Mean no. module genes recovered (of 6) 1

0 1% 5% 15% noise

Figure 3-5: Comparison of different methods in GBM mutation data.

A. Time complexity of SA, Greedy method and Exhaustive method, as compared to the increase of module size. B. Example of a simulated module (with coverage 50% and noise 5%). C. The average number of simulated 36

module genes recovered (out of the full 6 gene module) across 100 simulations. The SA method has better recovery than the greedy method, but both recover 5 of the 6 genes on average at 50% or more coverage. patients have a pattern as generated by the background model. Thus, the score of the module genes will be higher and the module will be more readily detected. At each coverage, the noise varies from low noise (on average 1% of the mutation statuses are flipped at random), to high noise (15% of the mutation statuses). We generate the module and the rest of the data 100 times for each setting of the parameters. Then, we assess the average number of genes from the gene set that is recovered by the algorithms, where 6 genes is the maximum (Figure 3-5C). Note that in each setting, including low coverage and high noise, at least three of the six module genes are recovered.

3.1.3.3 Application of Greedy GAMToC to TCGA GBM samples First, we explore modules of different sizes using only the mutation data, which is much more sparse than copy number data. The resulting mutation matrix contains 256 genes that are mutated in at least 2% of 283 patients with whole exome sequencing. For a module of size three, the simulated annealing method and the greedy method arrive at the same module of mutated genes.

Comparing this against the exhaustive method, we find that GAMToC recovers the best module in the data. When module size equals four, it would take 3.5 days for the exhaustive method to search all modules (Figure 3-5B).

Notice that while total correlation increases according to the module size, it does not make sense to compare different size modules in terms of total correlation. We use the G-statistics to overcome this issue (refer to method for detail), and calculate p-values based on the chi-square distribution for all modules. We find that the five-gene module containing TP53, IDH1, ATRX,

RB1, and PTEN is the most large and significant one in this example (Figure 3-6). In fact, TP53,

IDH1, ATRX, RB1 are all significantly positively correlated with each other. PTEN has a significant negative correlation with IDH1, as well as a positive correlation with mutation in RB1.

1.00e−15

IDH1 IDH1 IDH1 IDH1 1.00e−10 PTEN PTEN ATRX ATRX ATRX ATRX value − PIK3R1 TP53 TP53 TP53 TP53

TC p 1.00e−05

RB1 RB1 RB1

3 4 5 6 module size

Figure 3-6: Recovery of different module sizes in only mutation data.

The p-value associated with the total correlation is indicated on the y-axis, and the modules for each size are shown. For each size, the same module was found from the greedy and SA methods. Blue edges represent negative correlations between the genes, while red edges are positively correlated. Edge thickness denotes the strength of the association. Node size represents the frequency of alteration. Node border width represents the number of nonsynonymous mutations in that gene.

Next, we apply the greedy algorithm to a set of 273 tumors from the TCGA GBM project that have available copy number and exome sequence. Collating these data results in a mutation matrix of 756 alterations on the 273 samples. The greedy module recovered displays an interesting pattern of pairwise co-occurrence and mutual exclusivity between mutations (Figure

3-7A). It is important to note that total correlation finds a multi-gene structure of related alterations: as in the "exclusive or" example (Figure 3-4C), there may not be any strong pairwise relationships in a strong module. However, for visualization purposes we display the resulting modules in terms of their network of pairwise positive correlations (co-occurrence of a pair of genes) and negative correlations (mutually exclusive mutations). Thus, for the remainder of this work we provide a pair-based network visualization of the module structure.

A B

SPTA1 CDKN2A (region)

RB1 CDK4 (region) CDKN2A (region) BRSK2 EGFR TP53 RB1

BRSK2

IDH1 TP53

PAPLN

ADARB2 ATRX TMCO5A

C D TCOM5A BRSK2 RB1 CDKN2A (region) CDKN2A (region) EGFR ADARB2 KNAIN2 CD33 PTPN21 PTPN21 RB1 BRSK2 TP53 TP53 -log10p-value ATRX IDH1 TMCO5A CD33 Neural G-CIMP G-CIMP Classical Proneural NKAIN2 Mesenchymal

Figure 3-7: Networks of total correlation modules.

The legend is the same as in Figure 3-6, except for that node color represents average copy number amplification (red) or deletion (blue). A. The greedy module from glioblastoma. B. The Seed-GAMToC module, seeded with CDK4. C. The SA module. D. The genes from the greedy and SA modules are compared to subtypes of glioblastoma. The darker the red, the stronger the association (Fisher's exact test) of gene mutation status and that subtype.

We grow a greedy module up to the maximum feasible size, which is eight genes. In the greedy module, patients appear more likely to display mutations that co-occur with TP53, IDH1, and

RB1, or that are mutually exclusive with these genes. Patients with mutation or deletion of TP53 are significantly more likely to also have mutations in IDH1 and ATRX, and ATRX and IDH1 as a pair have the highest mutual information in the data set. The deleted and mutated gene RB1 strongly co-occurs with TP53 lesions, though it has no positive correlation with IDH1 or ATRX.

Deletion to the terminal section of chromosome 11p, which GISTIC2(Mermel et al. 2011) identifies as peak gene BRSK2, also frequently co-occurs with lesions of TP53 and RB1. The

11p15 region is imprinted, and it is known to be deleted, to undergo loss of heterozygosity, and to 39

have differential epigenetic regulation in multiple cancer types (Schwienbacher et al. 2000;

Onyango and Feinberg 2011).

Many of the genes that co-occur with TP53 alteration have a mutually exclusive pairwise relationship with copy number alterations in EGFR, CDKN2A region, or chromosome 10 deletion. The dominant effect of chromosome 10 deletion is likely the inactivation of the tumor suppressor PTEN, which is one of the most prevalent events across tumors. However, it is interesting that a large section of the chromosome is deleted, and not just PTEN. The greedy

GAMToC selects the GISTIC2 deletion peak on the terminus of 10p, containing ADARB2, as well as IDI1, IDI2, and WDR37. Very importantly, this region has stronger pattern of positive correlation with EGFR deletion, and negative correlation with IDH1 mutation, than does PTEN deletion, explaining its selection by the greedy method. While the full module of eight genes is very interesting, the seven gene module (removing CDKN2A region) is more statistically significant.

3.1.3.4 Seeding the greedy algorithm The greedy method has a disadvantage of performing only a local search for a high scoring module. It starts from the pair of genes with highest mutual information (pairwise total correlation), and uses a greedy approach to find a module that contains that pair. While we also develop the SA method to find other modules, the greedy method has two advantages for understanding cancer evolution. First, exploring the search space around the pair of genes with the highest mutual information is informative of processes in cancer, as we show above. Second, the greedy algorithm allows us to choose the starting point of the module search, by fixing an initial gene, that we call a seed gene. In this procedure, termed Seed-GAMToC, we identify a local maximum of total correlation that includes that seed gene. First, we find the partner gene for the seed gene. forming a gene pair with the highest mutual information, and we grow the greedy

module from this pair. Thus, we seek to characterize a given gene by finding what module of high total correlation contains that gene, or, in other words, the genetic context in which mutations of that gene appear. Discovering these relationships, such as the genetic context in which disruption of a query gene is advantageous, can illuminate the function of putative cancer genes.

Among the results of cancer genomics studies are frequent mutations in genes with a role in the cancer of interest that is not fully characterized. We run Seed-GAMToC for a number of genes that are significantly mutated or in copy number peaks in GBM patients, but were not selected by the greedy algorithm. We were interested in CDK4 because it is a cell cycle kinase that is focally amplified in GBM, and mutual exclusivity has been observed between amplification of CDK4, deletion to the CDKN2A locus, and deletions and mutations to RB1. We wondered what factors influence this mutual exclusivity, and we ran Seed-GAMToC starting from CDK4 (Figure 3-7B).

In fact, while CDKN2A is mutually exclusive with both CDK4 and RB1, the latter as a pair are not strongly mutually exclusive (chi-square p-value = 0.39). However, in patients with no CDKN2A deletions, their conditional mutual exclusivity is significant (chi-square p-value = 4x10-4). It is interesting that both CDK4 and RB1 have strong co-occurrence with other genes that are also mutually exclusive with CDKN2A. CDK4 co-occurs in patients with mutation to SPTA1, a recurrently mutated member of the spectrin cell scaffolding complex. Mutation to SPTA1 could impact cell adhesion, and mutations to other spectrins have been shown to affect cell cycle regulation(Metral et al. 2009). On the other hand, RB1 co-occurs with TP53 and its correlated genes. CDKN2A can regulate CDK4 and RB1, as well as TP53, explaining this discovery.

Because RB1, CDK4, CDKN2A all have roles in cell cycle, we also looked at the patterns associated with other significantly mutated cell cycle genes. For example, CDK6 plays a similar 41

role in promoting cell cycle progression as CDK4, and, like CDK4, this gene is strongly amplified. Seeding with CDK6, we find a strong correlation with PTEN deletion, and anti- correlation with ATRX and IDH1 mutation (Figure 3-8A). Thus, unlike CDK4, CDK6 may be a beneficial amplification in the context of the mitogenic PI3-kinase pathway, which is deregulated by PTEN deletion or mutation. On the other hand, another mitogenic event, amplification of

PIK3C2B (along with its chromosomal neighbor MDM2), seems to cooperate with deletion of

RB1 and amplification of the cell cycle promoting amplification MYCN (Figure 3-8C). One final gene closely related to cell cycle regulation is CCNE1, and amplification of this gene is strongly mutually exclusive with TP53 alteration(Figure 3-8B). One effect of TP53 inactivation is in fact de-repression of CCNE1, and CCNE1 likewise can mediate genetic instability(Hwang and

Clurman 2005). Thus, the module identified by the greedy method is useful for understanding the role of a query gene in glioblastoma development, including closely functionally related genes.

A B PAPLN

RB1 LSAMP CCNE1 BRSK2 RB1 BRSK2 CDK6 TP53

TMCO5A TP53 TMCO5A IDH1

PTEN ATRX NKAIN2

MYCN

GABRA4 PIK3C2B

RB1 KIF16B BRSK2

TP53

Figure 3-8: Networks seeded with query genes.

A. CDK6 seed. B. CCNE1 seed. C. PIK3C2B seed (co-amplified with MDM2).

3.1.3.5 Simulated Annealing results consistently identify a high-scoring module The SA algorithm provides an alternate mode of selecting a module, allowing us to more broadly search for a high scoring module. Unlike the greedy method, SA can escape local maxima and find a higher scoring module. Over the course of the semi-random sampling, the SA undergoes

“annealing”, becoming more selective for high total correlation modules. A run of SA will eventually converge on one module, but in practical settings SA will converge on a local optimum. Because there are many more copy number events than nucleotide mutation events, and all alterations are counted equally in GAMToC, the SA is more likely to converge on states involving broader copy number changes, making it somewhat less sensitive to mutational patterns or very focal SCNAs than the greedy algorithm. In multiple runs of the SA, one best module was found, that has a higher total correlation score than the greedy module (1.28 as opposed to 1.03), and is extremely statistically significant.

In the SA’s best module, a pattern appears that is related to that of the greedy module, but dominated by copy number changes (Figure 3-7C). As with the greedy module, the SA module has a set of genes that co-occur with mutation of TP53. This includes, as before, RB1 and BRSK2.

Additionally, deletion in chromosome 15, in GISTIC2 peak gene TMCO5A co-occurs with these genes, while another deletion region on chromosome 14 centered on PTPN21 is also associated with some of TP53's co-occuring partners. Mutually exclusive with TP53 and RB1 mutations is again deletion to the CDKN2A/CDKN2B locus.

3.1.4 Discussion Our algorithms search for genes with related occurrence of alteration across tumor samples, based on the premise that the joint alteration status of genes in tumor samples can inform us of the evolutionary process behind the cancer. Unlike mutual exclusivity methods that impose a single structure on the data, our approach is able to form a more comprehensive picture of alteration patterns that exist in cancer data. The result of applying GAMToC to the TCGA GBM data is a

network of genes with a jointly related mutation pattern, suggesting that the alterations in GBM do in fact follow an underlying structure. The interpretation of the module can be more complex, as opposed to mutual exclusivity, which is often interpreted as representing alternative mutations in a pathway. But one interpretation is that the co-occurring sets of gene lesions represent alternative pathways to glioblastoma development: there are different contexts in which these different lesions provide a selective advantage.

The interpretation of the sub-module structure as indicating routes to GBM development suggests that patients harboring different sets of mutations may have different characteristics. In fact, this pattern has been observed in the TCGA GBM cohort. Subtypes of glioblastoma have been identified by expression (Verhaak et al. 2010), as well as by methylation(Noushmehr et al. 2010), and these have been related to specific genetic alterations(Brennan et al. 2013). Patients with a methylation profile known as Glioblastoma CpG Island Methylator Phenotype (G-CIMP) have better survival, while patients with a gene expression pattern that follows the Proneural subgroup have different response to therapy. To support the hypothesis that the GAMToC module is indicative of these types of tumors, we examine if the GAMToC modules are related to these patient subtypes. We test whether patients with mutations to each module gene are more likely to fall into one of the subtypes. In result, the Classical and Proneural gene expression subtypes are strongly associated with certain module genes, as is the G-CIMP methylation group (Figure

3-7D). Thus, our approach successfully captures biological differences between patient groups, as reflected in different patterns of genetic lesions.

The Classical subtype typically has co-occurring mutations in EGFR and CDKN2A. Mouse models have suggested that activation of EGFR can cooperate with loss of the CDKN2A locus and PTEN to generate gliomas with high resemblance to GBM (Zhu et al. 2009). However,

rather than PTEN, the chromosome 10 deletions of ADARB2 are selected by GAMToC. This region is strongly co-deleted with PTEN (chi-squared p-value = 3.7x10-25), since in many cases of

PTEN deletion most of chromosome 10 is deleted. However, ADARB2, IDI1, IDI2, and WD47 have a stronger pairwise pattern with the other module genes chosen by GAMToC. Additionally, patients with this deletion are significantly more likely to fall into the Classical expression subtype (chi-squared p-value = .029), while PTEN is weakly associated with the Mesenchymal subtype (p-value = .086). Thus EGFR amplification, chromosome 9 deletion of CDKN2A and

CDKN2B, and ADARB2 locus deletion (including IDI1, IDI2, and WDR37) are all negatively correlated with TP53 and are all associated with the Classical expression profile.

In contrast to the better understood Classical subtype of GBM, the IDH1-p53 network associated with G-CIMP and with Proneural groups has been long studied but has so far remained of uncertain significance for tumor initiation in the brain. The strong co-occurrence of TP53 alterations with deletions of 11p15 (BRSK2) and 15q14 (TMCO5A) is an exciting novel finding.

While TP53, IDH1, ATRX, and BRSK2 are all highly associated with G-CIMP, TP53 and BRSK2 are also strongly associated with Proneural status. BRSK2 is particularly intriguing because it is a kinase that is highly expressed in brain and may be involved in apoptotic stress response(Y.

Wang et al. 2012) and cell cycle regulation(R. Li et al. 2012). Proneural tumors are also strongly associated with TMCO5A deletion, a lesion that, distinctively, is not associated with G-CIMP tumors. The genes in these regions may provide the missing element to recapitulate the gliomagenic process in these tumors.

It is also interesting to compare our modules with modules of mutually exclusive genes. Methods to find patterns of mutual exclusivity, such as MeMo (Ciriello et al. 2011) or DENDRIX(Mark D.

M. Leiserson et al. 2013), have pointed out genes also selected by GAMToC. These methods

sometimes claim to find new pathway interactions in this manner, exemplified by the mutual exclusivity between CDKN2A, CDK4, and RB1, or between CDKN2A and TP53. But GAMToC’s ability to find other relationships between mutations shows that the mutual exclusivity is related to the subtype-specific nature of mutations. It is very interesting to focus on the example of the retinoblastoma pathway, which can integrate signals from the mitogenic pathways (PI3-kinases,

PTEN), and DNA damage (TP53), among others. We find that mutations to the DNA damage

(TP53), cell cycle (RB1), and mitogenic (PTEN) pathways are prevalent across the glioblastomas, but that different specific alterations seem to confer subtle advantages in different mutational backgrounds. In Figure 3-9 we outline the subtype associations of genetic alterations affecting these pathways. For example, TP53 and RB1, as well as CDK4, are advantageous for G-CIMP and proneural tumors, while CDKN2A is a dominant lesion in classical glioblastomas, and

CCNE1, and CDK6 also occur less frequently in the Proneural tumors. Highly functionally related genetic alterations have been suggested to have similar effects. In the case of CDKN2A

(p16) and TP53, both lesions alter DNA damage response, while cell cycle regulation is transformed by mutations to CDKN2A, CDK4, CDK6, CCNE1 and RB1. However, far from the simplifying assumption that mutually exclusive events represent alternative equivalent routes to cancer development, clearly there are subtleties resulting in subtype-specific mutations. The data imply that mutations to genes in the same pathway are not in fact interchangeable.

A' !log10'p!value'

B' !log10'p!value' !log10'p!value'

C' CDKN2A' Proneural' Classical' p14' p16' Proneural/G!CIMP' TP53' PTEN'

MYCN' CDK4' PIK3C2B' (MDM2)' CDK6' CCNE1' RB1'

Figure 3-9 Cell cycle, DNA damage, and mitogenic gene subtype associations.

A. Fisher's exact test is used to assess association of mutation status with each subtypes. Darker red is stronger association. B. In reverse, this shows the depletion of mutation status of the gene in each subtype. C. A schematic of the functional relationships between genes involved in RB1 regulation. The fill of the boxes represents prevalent amplification or deletion of that gene in glioblastoma. The line on the outside of the boxes represents the subtype specificity of the gene, as calculated for part A. 47

More generally, our results also provide insight into the nature of subtype-specific lesions. As the method will detect any non-random pattern of alteration in a collection of samples, the resulting module may contain genes that are co-mutated because they are both present in tumors of the same subtype or environmental condition, rather than because of any direct functional interaction.

While patterns of joint lesion status do not allow us to distinguish between these two conditions, our results show that genetic context has a strong influence on selection. Thus, the distinction between subtype-specific co-alteration versus synergistic co-alteration may be thought of as a matter of the degree of selective advantage, rather than as two different phenomena. In conclusion, we have developed a method to uncover novel relationships between genes that are key to cancer development, and we have related the findings to previous subtypes of glioblastoma. Understanding the combination of genetic alterations present in patients with a tumor will help to target therapies to their pattern of aberrations. This application is an example of the power of a generalized entropy-based approach to gene set recovery.

3.2 GAMToC-L: Using patterns of co-selection of cancer genes to

identify and contextualize novel drivers

The results from GAMToC (3.1) were highly encouraging: genes such as BRSK2, that have not been highlighted in work in glioblastoma, stood out in this analysis. This gene is recurrently deleted—GAMToC relies on recurrently altered genes for input to the analysis. However, many genes are recurrently altered in glioblastoma, but very few have such a strong pattern of joint co- occurrence and mutual exclusivity. Therefore, the method is able to highlight potential driver genes with an evident pattern of selection that is not reliant only on recurrence.

We wondered if this idea could be extended: perhaps rather than starting from a list of recurrently mutated genes and finding those among them that have a non-random relationship, we could start from all copy number and point mutations, and use the total correlation to both identify driver genes and to find a module pattern. Thus far, most studies examining relationships between genes have started from the recurrent mutations because these represent the most likely driver genes

(Ciriello et al. 2011; Ciriello et al. 2013; Vandin, Upfal, and Raphael 2012; Akavia et al. 2010).

Because every cancer is different, it is well known that some low-frequency drivers will never be captured by recurrence-based measures, even with large sample sizes(Lawrence et al. 2014; Mark

D M Leiserson et al. 2014; Torkamani and Schork 2009). Particularly, this has left copy number data as an under-utilized resource in cancer genomics. There are so many copy number alterations in a given tumor that they clearly cannot all be important events for the tumor. But copy number changes are major events with a demonstrated high impact on gene expression and cellular function. Novel methods to find important changes, both in copy number and in nucleotide sequence changes, are strongly needed.

Total correlation could provide a new signal for positive selection in cancer. A module of genes cannot have a strongly non-random pattern when the component genes are extremely rarely mutated. However, presence of a strong module pattern that includes a gene could allow us to distinguish the likely drivers among genes that are altered at similar frequencies. In this section of my dissertation, I describe a new method, called GAMToC-L, (GAMToC-Landscape), that is able to explore the space of high total correlation modules and identify more subtle patterns of selection in a number of cancers.

This work is based off of the simulated annealing GAMToC method, with a few important changes. First, GAMToC only found modules among recurrent genes, limiting its search to 256

genes in the case of the TCGA glioblastoma data. GAMToC-L instead uses all genes with any copy number or point mutation, a set that comprises from 6000 to 12000 genes per tumor type.

Second, while GAMToC only considered one high scoring module, discovered through two different search methods, GAMToC-L examines the distribution of modules generated by the simulated annealing method.

I extend the results from GAMToC by applying the new method not only to glioblastoma, but also to lower grade glioma. This provides an interesting new contrast to the glioblastoma results, as lower grade gliomas share cells of origin with the higher grade glioblastoma, and the higher grade tumors sometimes arise from lower grade. I show that novel, but highly plausible, driver genes are uncovered, and patterns of co-mutated genes provide further insight into the biology of these cancers. This work is in preparation for submission.

3.2.1 Methods 3.2.1.1 Relationship with GAMToC As mentioned, GAMToC-L is heavily based off of the methods of GAMToC, as described in

3.1.2.3. Briefly, in GAMToC, a binary input matrix of samples by mutated genes is created. The first major difference between GAMToC and GAMToC-L is that in GAMToC-L the input is not restricted to recurrently mutated genes, but this matrix contains all genes that are mutated in more than f patients. As mentioned in the introduction, this increases the search set from 256 genes to

6120 genes mutated in three or more patients in the GBM data.

GAMToC’s simulated annealing procedure starts from a randomly generated module. It is important to note that we apply the same simplifying rule (see Figure 3-2) that allows only one gene per chromosome in the module. The module size is determined by the number of samples

available: with N samples, only log2(N) = M binary variables can possibly be observed in all of their states, so this represents the absolute upper limit on module size. In order to capture 50

complex relationships, we use this size for M. At each step of the simulated annealing, we randomly replace a gene from the current module, choosing a random replacement gene. This change is retained if it creates a higher total correlation, and otherwise it is probabilistically retained, and the probability is tuned by a temperature parameter t. Tests show that the initial temperature does not affect results, provided it is high enough (greater than .1). After a number of iterations, i, and a number of changes to the module, c, the temperature is lowered by a percentage, p. Over the course of the iterations, the temperature decreases and the total correlation increases to a plateau (Figure 3-10).

A) 0.2 2 B) 1 Total)correla-on) 0.1 0.8 0.15 1.5 0.05 0.6 0.01 0.1 1 0.005

accept 0.001 Greedy)TC) − 0.4 p 0.05 0.5 0.2 Temperature) 0 0 0 0 2 4 6 8 10 12 14 0 0.1 0.2 0.3 0.4 x 105 delta−TC Itera-on)

Figure 3-10 Effect of decreasing temperature

A. Over the iterations of the simulated annealing (x-axis), the temperature (blue line) decreases, and the total correlation (green line) increases, and its variance decreases. This example comes from the TCGA GBM cohort. The dashed line shows the total correlation attained by the greedy method, applied to the same data. B. Reproduced from 3.1.2.3, as temperature decreases the probability of accepting a change that lowers the score also decreases.

The simulated annealing will reach a local maximum, at a low temperature, defined by no change in u iterations. Then, if the maximum total correlation in the search space that was explored is not also the local maximum, the search will restart at the maximum previous value. The result of these iterations is a distribution in the space of modules. In GAMToC-L we use as much of this low temperature and high total correlation search space as is feasible. Thus, the distribution of modules over the module search space, a sort of metadata on patterns of mutation across tumor cohorts, becomes the input data for GAMToC-L. We call this metadata the module data.

3.2.1.2 Low-temperature module space appears to represent fluctuations around one solution The results shown in Figure 3-10A are illuminating for a number of reasons. As the temperature gets very low, the module space explored seems to narrow around a resulting total correlation score. This space seems to be distinct from that reached in the greedy search method: the total correlation remains higher than the greedy module total correlation throughout the final iterations.

It is intuitive that if a true module with at least M genes exists in the data, the module space explored will narrow to this true module: each step only changes one of the M genes, so any change that improves the total correlation score will improve it by selecting a gene with a relationship with the existing module genes. At the low temperature, any other gene will rarely be selected. Thus, GAMToC’s simulated annealing procedure will select a space of modules that are related to each other.

A gene frequencies B gene num partners 400 500

400 300

300 200 200

100 100

0 0 0 2 4 6 8 0.5 1 1.5 2 2.5 3 3.5 frequencies (log10)) number of partners (log10))

4 x 10 pair frequencies C 2

1.5

0.5

0 0 1 2 3 4 5 6 number of modules in which pair appears (log10))

Figure 3-11 Distribution of the frequency of genes and gene pairs appearing in the module data.

A. The log distribution of frequency of genes has an approximate normal distribution (kstest p = .97), indicating a highly skewed distribution of genes selected in the module data. B. The number of partners per gene, which is the number of other genes that appear in any module with a given gene, seems to follow an exponential distribution. C. The log distribution of the number of times each pair of genes appears.

Interestingly, less than one quarter of the genes included in the input data are present in any of the one million modules resulting from the final one million iterations of the simulated annealing, and these genes are chosen in a highly skewed distribution (distribution shown in Figure 3-11A).

Examination of how frequently pairs of genes are chosen together in the same module provides further support that GAMToC-L converges to a set of related genomic alterations. For each pair of genes we examine how many modules contain both genes of the pair (distribution in Figure

3-11C). The more frequently a pair of genes appears in the same module, the stronger is the total correlation of modules involving these genes. If more than one module exists in the data, we would expect to see a cluster of gene pairs that are frequently co-selected together. We create a visualization of the gene pairs in the GBM data in Figure 3-12. The visualization indicates that generally, some genes are chosen more in modules with many other genes. With a few interesting exceptions, frequently chosen genes co-occur with a broad set of other genes in the module data.

TP53

1 17 6 5 14 22 15 20 11 9 13 19 4 12

Figure 3-12 Frequency of co-selection of pairs of genes in the module data.

Each row and column is one gene, arranged in chromosomal coordinate order. Chromosome numbers are labeled on the x and y axes, and chromosomes are ordered for visual clarity. Columns containing genes on the same chromosome are outlined in the colored rectangles. Only chromosomes and genes selected in the low-temperature module data are shown. Each point is a pair of genes, with the darkness of the point showing how frequently the pair is co-selected in the module data. For example, TP53 (arrow) is the only gene on chromosome 17 chosen, and it is chosen frequently with almost every other gene in the module data. This can be viewed on the figure as the dark vertical and horizontal stripe of densely packed points indicating all genes that TP53 is chosen with (zoom in for best view). Note the lack of pairs of genes chosen together within a chromosome, which is by design.

3.2.1.3 Identification of consistently and recurrently selected genes We develop a method termed window-validation to identify the recurrently selected genes consistently chosen across iterations. Like cross-validation, the window-validation compares results in different subsets of a data set. In this case, the data is the module data resulting from the simulated annealing. Across the low-temperature iterations, the module space explored will vary: modules in nearby iterations will be more similar to each other than modules in more distant iterations. Thus, we use a sliding window approach to split up the data. Each window is a subset of consecutive iterations, and the makeup of the modules in the window is compared to the makeup found in the rest of the data.

For a given subset of the module data, for each chromosome, we identify the genes on the chromosome that are chosen at an elevated rate as follows. Let t represent the number of times a module in the subset includes a gene from the chromosome, and x represent the total number genes from that chromosome selected in the subset. Each gene on a chromosome is chosen mutually exclusively, by design, so the expected distribution of number of times each gene will be selected, given that a module contains a gene from the chromosome, is multinomial, with a uniform probability of 1/x for each gene. Thus for an individual gene, its probability of being chosen versus not being chosen would then be binomial with the same probability. We identify a the 95-percentile of the binomial distribution with number of trials t, and probability 1/x, in order to identify a cutoff for the number of times a gene is chosen that is more than expected.

We apply this procedure to the subset of the data in the window, and to the subset of the data not in the window, and we test whether there is a significant overlap in the genes chosen. If so, this indicates that the same genes are consistently and recurrently identified in the simulated annealing. This procedure is applied across all sliding windows. The genes (or localized

chromosomal regions, generally) that are consistently and recurrently selected across the sliding windows form our resulting module.

3.2.1.4 Iterative module selection We observe that a number of our greedy module genes from GAMToC are not selected in

GAMToC-L. Indeed chromosome 7 (EGFR) and chromosome 10 (PTEN, ADARB2) are never chosen in the low-temperature iterations. We have applied an iterative approach to finding multiple modules of interest in the data. All genes that were chosen in the low-temperature iterations are removed from the data. Additionally, any copy number alterations that are on the same chromosome as these genes are removed. This will prevent the same relationships from being re-discovered. Next, the procedure is re-run with the remaining genes. The subsequently selected module genes may have included strong interactions with the genes that were removed, limiting the relationships that can be discovered in this fashion. However, this procedure allows us to find more patterns in the data.

3.2.2 Results 3.2.2.1 Results in glioblastoma data The glioblastoma data provide an interesting result and point of comparison between GAMToC and GAMToC-L. As mentioned in the Method, we find that in the top module of GAMToC-L, the total correlation is substantially higher than the GAMToC greedy module. In fact, half of the genes from the greedy module are not present in GAMToC-L’s top module. In partial agreement with the GAMToC greedy results, we find two subsets of genes with a mutually exclusive relationship (Figure 3-13). One set of genes is associated with TP53 and RB1, while the other set is anticorrelated with these genes. Results include some genes identified via recurrent copy number alterations or point mutations, as well as some genes that are not recurrent enough to be identified on their own.

@1q25

NAA15 FBXW7 @4

BRSK2 @11 OSBPL8 @12

TP53 @17

@20q CDKN2A @9

LCTL MAP2K5 PIAS1 @15 DIS3 RB1 @13 @19p

GPR132 RCOR1 BCL11B CDCA4 @14

@6q15

Figure 3-13 GAMToC-L module for the GBM data.

Legend: node size represents how frequently the locus is chosen in the module space. Edge transparency also represents how frequently a pair is chosen. The width of the edge represents the strength of mutual information between a pair. Red edges are positively correlated, while blue edges are anticorrelated pairs. The color of a node represents its level of amplification or deletion in the cohort. The node border (here, only visible on TP53) represents the number of point mutations in the node.

In the TP53-associated group are a diverse set of co-occurring deletions. One of the most frequently chosen is the deletion in chromosome 11. This locus contains the recurrent deletion of

BRSK2: GAMToC identified this as a TP53-associated event, and the same region is also selected in GAMToC-L. Interestingly, GAMToC-L’s most recurrent selection in chromosome 11 is

BRSK2, indicating that the deletion identified by GISTIC2 is also the deletion with the strongest total correlation pattern. As mentioned in 3.1.4, the brain specific kinase BRSK2 is expressed in brain and may be involved in apoptosis as well as cell cycle regulation (Y. Wang et al. 2012; R.

Li et al. 2012). The module pattern suggests that one of the main effects of alteration in this kinase in glioblastoma may be its collaboration with TP53 and RB1. Chromosome 13 deletions identified by GAMToC-L include both RB1, which is known to co-occur in Proneural type glioblastomas with TP53 mutations, as well as DIS3, an exonuclease.

In contrast to this consistency with GAMToC, GAMToC-L finds different regions on chromosome 15 and on chromosome 14 from the regions selected in GAMToC’s simulated annealing result. As GAMToC only used the recurrent alterations as input, and these recurrent deletions are highly linked to the genes chosen, the recurrent mutations may have only been chosen in GAMToC for their linkage to these module members. The chromosome 15q deletion, identified by GAMToC as TMCO5A, is not chosen by GAMToC-L for the module. Instead

GAMToC-L chooses a region containing PIAS1, the protein inhibitor of activated STAT1, LCTL, and MAP2K5. The deletion that GAMToC selected on chromosome 14 was PTPN21. But

GAMToC-L selects others in the broad peak containing that gene, particularly CDCA4, and immediately 3’ of CDCA4, a G-protein coupled receptor, GPR132. The gene CDCA4 has been shown to repress E2F in regulation of cell proliferation(Hayashi et al. 2006).

A number of other copy number alterations also appear to co-occur in the TP53 deleted samples.

A deletion in chromosome 12 containing OSBPL8, Oxysterol binding protein-like 8, is not significantly recurrent. But GAMToC-L selects it, rather than its recurrently amplified neighbors

on 12q, because of its particularly strong relation with chromosome 11 deletion and TP53 alteration. Although this protein is not well studied in relation to cancer, this alteration may influence the well known metabolic transformation in cancer cells known as the Warburg effect(DeBerardinis et al. 2008). In addition, the cholesterol binding activity of other members of the oxysterol binding protein family has been shown to function as a scaffolding protein influencing ERK phosphorylation, in a key cell signaling pathway(P.-Y. Wang, Weng, and

Anderson 2005). This is an example of a locus that is not identified by recurrence, but that shows a strong pattern of relationship with other alterations. Another example is found in chromosome

4. The loci identified there are also not recurrently deleted, but we find NAA15 and FBXW7 have strong relationships with TP53 and other TP53-co-ocurring genes. The protein N-terminal acetyltransferase NAA15 may regulate translation and apoptosis(Arnesen et al. 2006). As mentioned in 2.1.2, FBXW7 is a known tumor suppressor in other cancers, but deletions in glioblastoma have not been reported.

Deletions in chromosome 6, while not particularly correlated with TP53 alteration, have a correlation with the TP53-associated GAMToC-L genes. Another set of genes is consistently anticorrelated with the TP53 group. This set of alterations show some positive correlation amongst each other. This includes the deletion to the CDKN2A locus, and co-occurring amplifications in chromosome 19 and chromosome 20. A final locus chosen by GAMToC-L is chromosome 1 amplifications: these are associated with TP53 and other of TP53's companions, but also show positive correlation with the chromosome 20 amplifications. These represent an interesting exception to the overall pattern of two mutually exclusive sets of genetic alterations.

When the results from the first iteration of GAMToC-L are removed, and the algorithm is re-run, a module similar to the greedy module from GAMToC appears (Figure 3-14), containing co-

occurring ATRX and IDH1 mutations, mutually exclusive with co-occurring chromosome 10 deletion and EGFR amplification. Some additions include PIK3R1, which is known to co-occur in IDH1-bearing glioblastomas, along with amplification of 8q24, near MYC, that occurs more in the IDH1 mutant samples.

@16p

HSPA13 ITSN1 NRIP1 @*21

ATRX @^X EGFR @^7

PIK3R1 @^5

IDH1 @^2 GTPBP4,LARP4B @^10

@8q24

Figure 3-14: Second module from GBM data. For legend see Figure 3-13.

3.2.2.2 Results from lower grade gliomas Lower grade gliomas (LGG) represent an interesting point of comparison with glioblastoma.

Glioblastomas often arise from lower grade gliomas, and all gliomas generally arise from glial cells. Like glioblastoma, lower grade gliomas have a number of subtypes, linked to distinct phenotypic outcomes. This is reflected in the top LGG module (Figure 3-15). As is well known, a strongly co-occurring deletion in 1p and 19q mutation is mutually exclusive with TP53 and

ATRX mutation. Both of these mutually exclusive genomic subtypes frequently have mutation in

IDH1. This pattern is reflected by the module chosen. Additionally, EGFR amplification and chromosome 10 deletion, lesions similar to the worse-prognosis classical subtype of glioblastoma, appear mutually exclusive with the IDH1-co-occurring alterations, much like in glioblastoma. Some nuance is added to this pattern. Again in resemblance to glioblastoma, chromosome 11 deletions co-occur with TP53 deletion in lower grade gliomas. The genes chosen include not only BRSK2, but also CDKN1C, a cell cycle regulator. Also co-occurring with the

TP53 group are amplifications affecting DEPTOR, also located near MYC. While MYC amplification can promote cell cycling, DEPTOR amplification is expected to have a different oncogenic effect, promoting Akt activation and inhibiting apoptosis (Pei et al.). Deletions containing NAA15, FBXW7, and other genes in 4q31 seem to co-occur more in the 1p19q cases, mostly lower grade oligodendrogliomas. A final region that is very interesting is SFI1, which does not fall into one of the three subtypes. It is mutually exclusive with the 1p19q cases, and co- occurs with both 11p15 deletion and the chromosome 7 and 10 copy number changes. This gene appears to be relevant in chromosomal segregation and thus may regulate cell cycle in a variety of contexts.

GGT5 SFI1 IDH1 @2 SLC5A4 @*22

@11p15 EGFR @7

TP53 @17

ATRX @X @10

DEPTOR @8

CDKN2A @9

@19q @1p

@4q31

Figure 3-15 Module for lower grade glioma. For legend see Figure 3-13.

3.2.3 Discussion Our new method has demonstrated power to explore the landscape of high total correlation gene modules and find those combinations of genes that have a shared non-random pattern of alteration. We have turned a limitation of copy number data, the linkage between genes on the

same chromosome, into an advantage. The simulated annealing process, run over many iterations, is an effective competitive procedure in which genes from the same chromosome are more likely to be retained in modules if they have a stronger total correlation with other module genes. This allows us to prioritize important copy number changes that may be drivers, and to identify important genetic alterations without using recurrence.

A few improvements could be envisioned for this approach. In particular, it will be interesting to further explore the differences in modules containing different genes from the same chromosome.

Genes from the same chromosome will usually have a strong correlation with the same genes, but more subtle patterns may exist. For example FBXW7 and NAA15 deletions, both on chromosome

4, are both highly correlated with TP53 and RB1 alterations, but there may be some distinction between the modules containing each of these genes. Methods to dissect these patterns at the sub- chromosomal level will provide further insight into driver alterations and their strongest relationships. Another improvement could be some mechanism to weight point mutations such that they appear in the module data at a similar frequency as copy number changes.

Finally, it will be interesting to apply this method to skin melanoma. Melanoma does not have the distinctive subtype pattern present in the gliomas, so we would not expect the same pattern of mutually exclusive sub-modules of genes. Thus, as expected, preliminary results from

GAMToC-L show subsets of co-occurring genetic alterations. As melanoma undergoes a high rate of genetic damage, a method that can sort through the many passenger alterations and find subsets of cooperating driver genes will be of high interest, even if distinctive molecular subtypes are absent.

4 Genetic similarity between cancers and comorbid Mendelian diseases identifies candidate driver genes Despite large-scale cancer genomics studies that uncover key somatic mutations driving cancer, most studies, and much of my dissertation work, focus only on patterns of genomic alterations in tumors. In this section I propose that analysis of comorbidities of Mendelian diseases with cancers provides a novel, systematic way to discover new cancer genes. If germline genetic variation in Mendelian loci predisposes bearers to common cancers, the same loci may harbor cancer-associated somatic variation. Compilations of clinical records spanning over 100 million patients provide an unprecedented opportunity to assess clinical associations between Mendelian diseases and cancers. I systematically compare these comorbidities against recurrent somatic mutations from more than five thousand patients across many cancers. Using multiple metrics for genetic similarity, I show that a Mendelian disease and comorbid cancer are indeed have genetic alterations of significant functional similarity. This result provides a basis to identify candidate drivers in cancers including melanoma and glioblastoma. Some Mendelian diseases demonstrate

“pan-cancer” comorbidity and shared genetics across cancers. This work is under review at

Nature Communications.

4.1 Introduction

Recent years have brought an explosion in the number of genomically profiled tumors, and the promise of finding most genetic loci containing cancer-predisposing variation seems within reach.

While algorithms to sort through the complex landscape of tumor lesions(Lawrence et al. 2013;

Mermel et al. 2011) have revealed recurrently altered “driver loci” – those somatic or germline genetic defects that are most likely to trigger the disease – the directory of relevant genes and the catalogue of their roles in tumor progression remain incomplete. The search for cancer genes has expanded to additional informative patterns, such as mutual exclusivity of mutation across patients and functional relationships between cancer-altered genes(Ciriello et al. 2011; Vandin,

Upfal, and Raphael 2012; G. Wu, Feng, and Stein 2010).

One historical source of information on key cancer alterations may be found in Mendelian disorders, rare conditions that have long provided insight into a wide array of human disease processes. Some of the first genes linked to cancer were characterized by their highly penetrant familial association with certain tumors. Studies of familial retinoblastoma led to the identification of RB1 as a tumor suppressor(Friend et al. 2014), while cases of Li-Fraumeni syndrome showed that germline mutation of TP53 pleiotropically predisposes patients to many cancers(Malkin et al. 1990). Other Mendelian disorders, such as Rubinstein-Taybi syndrome, involve a primary phenotype apparently unrelated to cancer, yet the bearers are known to have an increased tumor risk(R. W. Miller and Rubinstein 1995). Recent studies demonstrating that

Rubinstein-Taybi’s primary causative gene, CREBBP, is also recurrently somatically inactivated in a number of cancers(Pasqualucci et al. 2011; Kishimoto et al. 2005; Yang 2004; Mullighan et al. 2011) have provided a likely explanation for this comorbidity. These examples suggest that

Mendelian germline mutations could predispose Mendelian disease patients to common cancer by

disrupting cellular functions that in the majority of cancer patients are altered by somatic rather than germline genetic events.

Recently, Electronic Health Record (EHR) data sets of unprecedented size have provided statistical power to measure comorbidity of pairs of diseases(Blair et al., 2013; D.-S. Lee et al.,

2008; Park, Lee, Christakis, & Barabási, 2009). With the recent increase in the amount of data recorded in EHRs it is newly possible to detect clinical associations even in diverse rare diseases, such as some Mendelian diseases. These results have suggested that comorbidity is indicative of shared germline genetic architecture. Here, we propose that Mendelian disease comorbidity with cancer could be associated with a relationship between Mendelian disease loci and driver loci somatically altered in cancer. It is possible that genetic variants that cause Mendelian disease with high cancer comorbidity also provide a selective advantage to aberrant cells of a developing tumor, leading to this predisposition to a certain type of cancer. If this is correct, exactly the same

Mendelian loci and molecular pathways incorporating their products would be involved in a somatic context in tumors of patients lacking the germline mutation. Thus, comorbidity calculated from EHRs spanning large numbers of patients could provide a novel line of evidence for functional involvement of some genes as cancer drivers.

By integrating clinical data from more than 100 million patients with somatic genomic information from thousands of tumors from The Cancer Genome Atlas (TCGA)(“The Cancer

Genome Atlas”), we explore the connection between Mendelian diseases and common cancers.

First, we examine the hypothesis that comorbidity between Mendelian disease and cancer may be due to similarities between the genes involved in each. We find that comorbid diseases display statistically significant genetic similarity. Then, we use this relationship to test genetic similarity for comorbid pairs of Mendelian disease and cancer, identifying those disease pairs with shared

cellular processes. For each cancer, we prioritize these comorbid and genetically similar

Mendelian disease genes and pathways as candidate novel cancer drivers.

4.2 Comparing Mendelian disease and comorbid cancer

4.2.1 Integration of disease comorbidities and genes In the work of Blair, et al.(Blair et al. 2013) the authors estimated comorbidity for a set of diseases well characterized by patient billing codes, comprising 95 Mendelian diseases and 65 complex diseases, including 13 common cancers. Comorbidity was calculated using seven EHR datasets, including the MarketScan insurance claims data covering nearly 100 million patients.

For each complex disease, they compared its incidence in Mendelian disease patients against its marginal incidence. They crossed patient zip code information with US census data to obtain demographic, socioeconomic, and environmental factors. Then they corrected for these confounders, as well as for errors in billing codes, using a regression approach. Combining these analyses, they estimated relative risk for a complex disease in Mendelian disease patients, as well as a significance level. We use these estimates throughout this work. For each Mendelian disease billing code set, the authors curated a list of corresponding diseases, each linked to genetic loci(McKusick-Nathans Institute of Genetic Medicine, Johns Hopkins University (Baltimore).

We updated the mapping of diagnosis codes to genes using OMIM as well as OrphaNet data(Hoehndorf, Schofield, and Gkoutos 2013). Utilizing their work and other curation, we find a median of four genes related to each Mendelian disease type (the full distribution is shown in

Figure 4-1a).

Of the 13 cancer diagnosis code sets included in the Blair analysis, 10 correspond to one or more tumor types profiled in TCGA. These 10 diagnosis codes correspond to 15 TCGA tumor types, including melanoma, glioblastoma, and other common cancers, with genomic data across a total of 5,667 patients. We downloaded the calls of recurrently altered genes as assessed by the Broad

Institute and made available in the Firehose (http://www.broadinstitute.org/cancer/cga/Firehose) download data set of 9/23/2013. MutSigCV(Lawrence et al. 2013) assigns a statistic for evidence of selection for mutation of a gene across a set of tumors. For each tumor type, we select those genes with a q-value statistic less than .25. GISTIC2(Mermel et al. 2011) identifies genes in significantly recurrent and focal regions of copy number amplification or deletion, and we include only the genes in copy number peaks that contain fewer than 50 genes. Each tumor type has from zero to hundreds of associated genes either mutated or copy number altered. A median 155 genes are recurrently genetically altered per tumor type (Figure 4-1b). Entrez gene info data

(ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/GENE_INFO/Mammalia/Homo_sapiens.gene_info.gz ) was used to find common identifiers between all data sets

a 70

10 Num Mendelian disease groups

0 0 10 20 30 40 50 b Num genes per Mendelian disease group 5

2 Num cancers

0 0 50 100 150 200 250 300 350 400 Num genes per cancer

Figure 4-1 Distribution of number of genes per disease. a. The number of genes per Mendelian disease group ranges from 0 (for chromosomal disorders) to 51 (for retinitis pigmentosa), with a median of 4 genes per disease. b. The number of recurrent genes per cancer ranges from 11 for kidney chromophobe to 397 for lung adenocarcinoma, with a median of 155 genes.

As can be seen in Figure 4-1, some Mendelian diseases have multiple causal genes and the severity and rarity of Mendelian diseases also varies widely. We investigate factors related to the number of genes per Mendelian disease, and we find that this number is somewhat associated with severity and with rareness of the Mendelian disease, two factors that can influence the overall population of surviving adults with the disease (Figure 4-2a-b).

a Spearman rho = 0.10 p=0.026 b Spearman rho = 0.16 p=0.005 5 5

4.5 4.5

4 4

3.5 3.5

3 3

2.5 2.5

2 2

1.5 Age of death category (5 = normal) 1.5 Rareness category (5 = most common)

1 1 10 20 30 40 50 10 20 30 40 50 Number of genes in the associated disease Number of genes in the associated disease c Spearman rho = 0.25 p=0.02

4 # comorbid cancers per MD 2

0 10 20 30 40 50 # genes per MD Figure 4-2 Characteristics of Mendelian diseases a,b. Each point is a gene. For each gene we plot the number of other genes associated with the Mendelian disease category, against the rareness and severity are associated with the particular Mendelian disease variant as cataloged in Orpha database. For example, the gene ZIC2 is associated with holoprosencephaly, which has a total of 10 genes, and ZIC2's associated disease is annotated at a frequency of 1-9 / 100,000 (the second least rare category). It is important to note that this information is approximate and contains many missing values. Only genes with available information are plotted. c. Each point is a Mendelian disease, and the number of genes associated with a disease is significantly correlated with the number of cancer comorbidities

Most importantly for this study, we find that number of genes per Mendelian disease is correlated with cancer comorbidity ((Figure 4-2c). This has a number of possible explanations. One is that more rare, or more rarely diagnosed, diseases lack power to detect both causal genes and to detect comorbidities in clinical records. Another explanation is that Mendelian diseases with more genes annotated are more likely to have disease subtypes (one or more of these causal genes) that are related to cancer. In any case, any analysis of the comorbidity data must take this association into account. 70

4.2.2 Genetic similarity of comorbid diseases Next, we compare the sets of genes associated with a Mendelian disease to the recurrently genetically altered genes in TCGA. We consider multiple genetic similarity metrics, with the goal of assessing whether comorbidity is significantly related to shared genetics. The approach is outlinedFigure-1 in Figure 4-3a. (Rabadan) a b EHR$comorbidity$of$ diagnosis$codes$$ Map codes Mendelian Cancer Gene,enrichPathway BioGRID HumanNet Coexpr. Candidate Aromatic)amino)acid) to diseases metabolism)(pigment) SKCM ✔ ✔ Heart)&)Skeletal)) TCGA$ OMIM$ (Rubinstein=Taybi) SKCM ✔ ✔ Heart)&)Skeletal)) Gather genetically (Rubinstein=Taybi) GBM ✔ ✔ altered genes for Heart)&)Skeletal)) each disease (Rubinstein=Taybi) LGG ✔ ✔

Holoprosencephaly LGG ✔ ✔ ampliﬁed$ Germline mutated$ altered Holoprosencephaly GBM deleted$ Diamond=Blackfan GBM ✔ ✔

Genetic similarity of Diamond=Blackfan GBM ✔ ✔ comorbid diseases ...... Comorbidpairs Mendelian of disease and cancer

Gene enrichment Pathway enrichment

Coexpression Network connections Figure 4-3 Outline of the approach. a. Integration of the data and overview of genetic similarity metrics. b. Examples of comparison of pairs of diseases. All comorbid pairs of Mendelian disease and cancer with TCGA data are compared. Genetic similarity of comorbid diseases is assessed using multiple metrics. A simple combination of presence of any one of the genetic similarity metrics, after correcting for the number of comorbid pairs, is used to predict novel cancer driver loci.

Our similarity metrics are first evaluated on the aggregate of comorbid diseases in order to test the hypothesis that comorbidity is significantly related to shared genetic factors. Then, we use analogous tests for the pairs of diseases, in order to identify comorbid Mendelian disease and cancer with evidence of related gene sets. Below, we describe both uses of each metric.

In the first genetic similarity metric, we examine whether the same genes responsible for a

Mendelian disorder are more likely to be altered in comorbid cancers. For each of 427 pairs of comorbid Mendelian disease and TCGA cancer, we assess how many genes are shared (Figure

4-4). The gene enrichment metric scores the overlap of the Mendelian disease gene set of size m, within a cancer gene set of size c. The score assesses whether the number of genes in the overlap between the two sets is more than expected. For the per-pair score, we use a binomial model with success probability based on the fraction of all assayed genes that contain variants associated with the Mendelian disease, and number of trials corresponding to the cancer recurrently mutated gene set size c, and number of successes corresponding to the size of the overlap, v, between the sets:

! Binomial(v, c, ). # !"#"$

In all comorbid pairs, 41 genes are shared between the Mendelian causal gene set and the recurrently somatically altered cancer gene set. We test whether this number of genes shared across the 427 pairs of Mendelian diseases and comorbid cancers is more than would be expected at random. Our test uses a simulated convolution of the 427 binomial tests: for each pair, the binomial model, as before, has a success probability based on the fraction of total genes that are

Mendelian disease genes, and a number of trials based on the number of recurrent cancer genes.

Thus the convoluted distribution can be simulated as:

! BinomialSample(� , ! ). In other words, the samples from each comorbid !,!∈!"#"$%&'()&$ ! # !"#"$ pair are added to generate an expected distribution. The model is simulated 100,000 times to compare to the observed value. We find that 41 occurs in 2.1% of random trials (Figure 4-5a).

Lipoprotein Deficiencies Hereditary Sensory Neuropathy comorbid One gene shared Haemophilia 5 Genetic Anomalies of Leukocytes comorbidity Disorders of Straight Chain Amino Acid Metabolism Two genes shared Disorders of Phosphorous Metabolism undetected Chronic Granulomatous Disease Cerebral Degeneration Due to Generalized Lipidoses Long QT Syndrome Inherited Anomalies of the Skin Hereditary Hemorrhagic Telangiectasia 0 Degenerative Diseases of the Basal Ganglia 0 5 10 15 Combined Heart and Skeletal Defects Sickle Cell Anemia Hypopituitarism Dopa−Responsive Dystonia Thalassemia Spinocerebellar Ataxia Disorders of Urea Cycle Metabolism Severe Combined Immunodeficiency Disorders of Copper Metabolism Cystic Fibrosis Congenital Ichthyosis "Polycystic Kidney, Autosomal Dominant" Retinitis Pigmentosa Congenital Disorders of Purine/Pyrimidine Metabolism Circulating Enzyme Deficiencies Inherited Adrenogenital Disorders Glucose−6−Phosphate Dehydrogenase Deficiency Diamond−Blackfan Anemia Non−Specified Osteodystrophy "Pervasive, Specified Congenital Anomalies" Glycogenosis Friedreich Ataxia Specified Anomalies of the Musculoskeletal System Non−Specific Autosomal Deletion Syndromes Familial Dysautonomia Anophthalmos/Micropthalmos Systemic Primary Carnitine Deficiency Specific Nail Anomalies Osteogenesis Imperfecta Immunodeficiency with Increased IgM Huntington Disease Holoprosencephaly Facial and Skull Anomalies Erythromelalgia Congenital Hypogammaglobulinemia Congenital Hydrocephalus Congenital Hirschsprung Disease Chronic Progressive External Ophthalmoplegia Chondrodystrophy Sensory Retina Dystropies Disorders of Aromatic Amino Acid Metabolism Congenital Pigmentary Anomalies Congenital Ectodermal Dysplasia Androgen Insensitivity Syndrome LGG GBM KIRP KIRC KICH STAD LUAD BLCA LUSC BRCA PRAD READ UCEC COAD SKCM Figure 4-4 Genes shared in comorbid diseases

The genes shared in comorbid diseases is counted across all pairs. By comparing it to a null distribution based on number of Mendelian and cancer genes, we can assess if more genes are shared than expected. Cancer abbreviations are in from TCGA.

The pathway metric utilizes the NCI Pathway Interaction Database and the PharmGKB subsets of the Consensus Pathway Database (Kamburov et al. 2013) in order to obtain a diverse and non-

redundant set of pathways. The set contains 1343 pathways and a total of 4954 genes. We create a gene list containing the union of all genetically altered cancer genes across all of the cancers studied, and we remove all pathways with enrichment in this list in order to filter very general cancer cellular processes. We score strength of the overlap of a cancer gene set within each gene set associated with each remaining pathway using the same binomial gene enrichment score, then corrected by the number of pathways with the Benjamini-Hochberg method(Benjamini and

Hochberg 1995). Many pathways have no overlap with a cancer’s gene list, so the enrichment score for these is 1.0. For the Mendelian diseases, we consider a pathway to be affected if it contains any Mendelian disease gene. To assess the similarity for a pair of diseases, we use the

Spearman correlation coefficient of the pathway scores for each disease across all pathways, with the Spearman significance statistic providing our per-pair score.

For the aggregate score across comorbid pairs, we use a cutoff on cancer enrichment (q-value <

.1), and we count the number of out of the n pathways that are both enriched in the cancers (c), and involved in the Mendelian disease (m). We find 136 pathways shared in comorbid pairs. We assess whether this number of overlapping pathways is more than expected using the convolution of hypergeometrics, similar to the gene enrichment convolution:

!,!∈!"#"$%&'()&$ HypergeometricSample(�, �!, �!). The results are shown in Figure 4-5b. In order to ensure that the significance is not only due to two Mendelian disorders with the most pathways impacted, we also run this test when Rubinstein-Taybi syndrome and Pervasive

Specified Congenital Anomalies are removed: in this case only 81 pathways are shared but the overlap is still highly significant.

Our next test of genetic similarity between comorbid diseases uses well-studied gene interaction networks. The network metric measures the number of direct interactions of each Mendelian 74

disease gene set with the cancer gene set. This number is compared to the number found in a set of shuffled networks, created using a degree-preserving randomization algorithm(Maslov and

Sneppen 2002). In this randomization algorithm, a network is shuffled by repeatedly re-wiring pairs of edges, in order to preserve each node’s number of connections but randomize which genes are connected to each other. A pair of diseases is considered similar if fewer than 5% of random networks have the same or higher number of interactions. For the aggregate score, we count over the Mendelian diseases, the number of edges between a Mendelian disease's genes and the set of comorbid cancer genes. This count is compared against the count from the shuffled networks. We use two networks to independently score our disease pairs. In the BioGRID binary interaction data set(Stark et al. 2006), a curated set of genetic interations and protein interactions, there are 140,402 edges on 14,112 nodes, covering 86% of Mendelian disease genes and all but four of our Mendelian disease sets. In all, there are 797 direct edges between comorbid genes in this network, a number found in less than 2% of random networks (Figure 4-5c). Another network, HumanNet, is constructed by integrating a number of data sources(I. Lee et al. 2011).

HumanNet trains its integrated data set on Gene Ontology categories of genes, and it assigns a confidence score, in terms of log-likelihood of interactions, to each learned edge. We take the top

10% most confident edges, resulting in a network with 7,931 nodes and 47,934 edges. In

HumanNet, there are 296 direct edges between comorbid disease genes, which is a number found in only 0.2% of random networks (Figure 4-5d).

a Genes shared b Pathways shared 0.4 0.4

0.3 0.3

0.2 2.1% 0.2 0%

0.1 0.1 random distribution random distribution

0 0 10 20 30 40 50 40 60 80 100 120 140 number number

c BioGRID edges d HumanNet edges 0.4 0.25

0.2 0.3

0.15 0.2 1.7% 0.2% 0.1

0.1 random distribution random distribution 0.05

0 0 700 750 800 850 220 240 260 280 300 number number

Figure 4-5 Aggregate similarity of comorbid diseases a. The number of genes shared among pairs of comorbid diseases is 41, more than all but 2.1% of the generated null model (see Method for details about null model for gene and pathways shared). b. The number of pathways shared is 136. c, d. The number of edges shared between a Mendelian disease’s genes and the genes involved in comorbid cancers, shown for two different gene networks. In BioGRID, there are 797 edges, while in HumanNet 296 edges are found.

It is important to note that well-known Mendelian cancer syndromes were removed from the analysis before testing the association of comorbidity and genetic similarity. This includes: Li-

Fraumeni (TP53, CDKN2A), neurofibromatosis (NF1, NF2), Cowden syndrome and related hamartomas (PTEN, STK11), tuberous sclerosis (TSC1, TSC2), and dyskeratosis syndromes

(TERT and other genes involved in telomere maintenance). We do not include these known germline cancer genes in our analysis because we wish to assess the significance of novel

Mendelian disease associations with cancer. These cancer syndromes, as would be expected, are 76

each comorbid with multiple cancers, and they show many shared genes and pathways with the cancers (Supplementary Table 3).

Thus, using a number of lines of evidence, we have shown that the genes involved in Mendelian diseases have a specific functional relationship with the genes altered in co-occurring cancers, and most of these connections are novel. Therefore, comorbidity may be due to the genetic similarity relationship. In order to use comorbidity as a source of candidate novel drivers for each cancer, we use the per-pair scores of genetic similarity that we can apply to each pair of comorbid diseases. These per-pair metrics are related to the aggregate measures, as discussed above.

To these pairwise and aggregate measures of similarity, we wished to add an entirely unbiased source of information on functional similarity and cell-specific expression. We developed a coexpression metric utilizing the data from FANTOM5(Consortium, Pmi, and Dgt 2014). The

FANTOM5 data covers a diverse range of 889 cellular states, assessing promoter activity in each gene in each cell or tissue type. We download the human CAGE peak data quantified by transcripts per million

(http://fantom.gsc.riken.jp/5/datafiles/latest/extra/CAGE_peaks/hg19.cage_peak_tpm_an n.osc.txt.gz). Adding all peaks that are assigned to the same gene, we create an estimate of aggregate expression of each gene in each sample. As we wish to measure whether genes involved in a pair of diseases are expressed in the same conditions, we calculate coexpression of pairs of genes using the Pearson correlation coefficient. To calculate our coexpression similarity for a pair of Mendelian disease and cancer, we consider that significantly elevated coexpression between any cancer gene and a set of Mendelian disease genes represents interesting similarity.

Thus, for each cancer gene we compare whether the set of Mendelian disease genes has high coexpression with that cancer gene, as compared against the distribution of coexpression of all other genes with the cancer gene. We test this for each cancer gene using the Wilcoxon rank-sum

test. The p-values are then corrected for the number of cancer genes tested using the Benjamini-

Hochberg method.

We correct the genetic similarity scores for the number of comorbid diseases considered.

Coexpression has the most instances of similarity, most likely due to the fact that more genes can be compared and many types of functional relationships can be captured with coexpression. After correction, the gene enrichment and network metrics have very few instances of significant similarity, a point that is discussed below. Our candidates comprise the significantly functionally connected genes from comorbid and genetically similar disease pairs. The scores are shown in

Supplementary Table 3.

4.3 Mendelian disease comorbidity and cancer processes

4.3.1 Prediction of diseases with shared cellular processes From the per-pair genetic similarity metrics, we have generated a list of candidate linked

Mendelian diseases, and associated genes and processes, across 15 TCGA cancers. The complete resulting list of genes, and genetic similarity scores, associated with each linked disease pair is available in Supplementary Table 3. To provide examples and to demonstrate their relevance we highlight some candidates implicated for cutaneous melanoma and brain neoplasms.

Cutaneous melanoma is often located on sun-exposed sites, undergoing a high rate of genetic damage. Our findings can highlight both recurrently altered genes in melanoma and comorbid

Mendelian genes as potential cancer drivers. A central transcription factor involved in melanocyte cell fate, MITF, is related to multiple Mendelian diseases comorbid with melanoma. This gene has a complex role in this cancer: while it is recurrently amplified in 26% of TCGA melanomas, possibly promoting melanocyte proliferation, it is also frequently deleted (11% of cases).

Suppression of the gene is also advantageous for the growing cancer, as it reduces terminal 78

differentiation and senescence in melanocytes(Levy, Khaled, and Fisher 2006; Yajima et al.

2011). The melanocyte’s primary receptor MC1R, upstream of MITF, its other upstream activators, PAX3 and SOX10, as well as MITF’s key target, TYR, are all associated with

Mendelian disorders comorbid with melanoma (Figure 4-6).

MC1R% cAMP

PAX3% SLC45A2% SOX10% SOX2% OCA2% MITF% PAX6%

TYRP1% SNAI2% BCOR% Q79.8 incl. Waardenburg TYR% OTX2% Q11 microphthalmos E70.2/3 incl. albinism EP300% TP53% Q87.2 incl. CREBBP% Rubinstein-Taybi

Figure 4-6 Depiction of comorbid diseases with skin melanoma

Comorbid diseases are shown in terms of the roles of the Mendelian disease genes in the melanocyte development program as well as other cancer related processes. Genes that are recurrently somatically mutated in melanoma are highlighted. Solid edges represent interactions from the literature, while the dashed edges represent significant coexpression. Orange outlines represent genes with common polymorphisms conferring increased melanoma risk.

Of these, MC1R and TYR are associated with oculocutaneous albinism (included in International

Classification of Disease, revision 10 (ICD10) billing code E70.2/3, melanoma relative risk 95% confidence interval (CI) = (2.16 - 5.19)). MC1R is among the recurrently deleted genes in melanoma. Germline variants of MC1R, causing red hair, have been implicated as a risk factor for melanoma via both pigmentary and non-pigmentary pathways(Cao et al. 2013; Raimondi et al.

2008), and polymorphic variants of TYR, which leads to a green eyes, also confer significant, though lesser, risk(Gudbjartsson et al. 2008). Other albinism-related genes have significantly

elevated coexpression with MITF (p = .020) as well as MITF's target gene(Hoek et al. 2008)

KCNAB2 (p = 0.0093). KCNAB2 is recurrently deleted in the melanoma cases.

While the candidate melanoma genes associated with albinism are not recurrently genetically mutated in melanoma, we examine their patterns of expression for evidence of a functional contribution to the disease. We download Level 3 RNASeq data from TCGA portal, and the

RSEM(B. Li and Dewey 2011) expected counts are rounded to create the input to the analysis.

We transform these using the variance stabilizing transformation from DESeq2(Love, Huber, and

Anders 2014), which is recommended for clustering data. We then cluster melanoma tumors by their expression of these genes using consensus clustering methods implemented in

ConsensusClusterPlus(Wilkerson and Hayes 2010), and we find stable clusters (Figure 4-7a). An optimum clustering is found (based on change in classification consistency) of k=4. Three main large clusters are consistent through k=3 to k=6. We assess clinical outcome in these groupings, using the R package Survival(Therneau 2012) to assess survival difference between the groups and to plot, based on the available TCGA clinical data. Cluster assignments are highly predictive of patient survival (p = 0.0022, Figure 4-7b). This suggests that indeed this pathway is highly relevant for melanoma progression.

Color Key and Histogram 400 200 Count 0

5 10 15 Value

Consensus clusters a TYR

SLC45A2

TYRP1

HPS1

DTNBP1

HPS6

BLOC1S3

OCA2

FAH

HPD

TAT

HPS5

AP3B1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 Chisq= 14.6 on 3 degrees of freedom, p= 0.00215 b 1.0 0.8 0.6 survival 0.4 1 2 3 4 0.2 0.0

0 2000 4000 6000 8000 10000

days

Figure 4-7 Analysis of the role of albinism related genes in melanoma. a. Transformed expression of the genes in 471 melanoma samples is shown. Samples are arranged according to their consensus cluster tree, and genes are clustered using Pearson correlation coefficient. The consensus cluster assignments for k=4 are shown by the colored label at the top. b. Survival analysis for the four classes using the log rank test shows a significant distinction in prognosis among groupings assigned using only the expression of these genes

Also regulating MITF activity are its coactivators EP300 and CREBBP(Sato et al. 1997), genes associated with the melanoma-comorbid Rubinstein-Taybi syndrome (code group Q87.2, relative 81

risk 95% CI = (1.19 - 1.99)). EP300 is recurrently amplified (36% of the TCGA melanomas), but also frequently deleted (7% of cases). Rubinstein-Taybi shares many pathway enrichments in common with melanoma (Figure 4-8), including "melanocyte development and pigmentation pathway" and "Regulation of nuclear beta catenin signaling and target gene transcription", both of which involve MITF. The amplifications of EP300 are significantly more likely to co-occur in the same patients with MITF amplifications (one-tailed Fisher’s exact test, p = 0.0041), suggesting cooperation between the alterations, and a particular role for these genes in melanoma: the histone acetyltransferase activity of EP300 might enhance the function of an oncogenically amplified MITF. CREBBP and EP300 defects have also been linked to aberrant TP53 and BCL6 regulation in some lymphomas(Pasqualucci et al. 2011).

0.8 SKCM 0.6 Rubinstein−Taybi

0.4

0.2

Cancer enrichment ( − log10) 0 500 1000 Pathway index (ordered by SKCM enrichment) Figure 4-8 Pairwise pathway metric for Rubinstein-Taybi and melanoma

For a disease pair, the pathway metric compares the pathways impacted by the Mendelian disease to the pathways enriched for the cancer gene sets. Here, pathway enrichments for melanoma genes (blue) are compared to pathways involved in Rubinstein-Taybi syndrome. Each vertical red line represents one pathway impacted by a Rubinstein-Taybi gene. The pathways are sorted by their enrichment in melanoma. The Spearman correlation between the corrected p-values of melanoma and the impacted pathways of Rubinstein-Taybi for the pathways is -.25, p = 6.3x10-21.

Comorbidity of melanoma with ectodermal dysplasias (ICD10 code Q81, melanoma relative risk

95% CI = (6.01-17.84)) may highlight the importance of tissue invasion in melanoma progression. The ectodermal dysplasia disease epidermolysis bullosa can arise from genetic alteration to proteins involved in structural support, tissue integrity, and adhesion in the dermis 82

and epidermis. Although the chronic inflammation and tissue damage associated with epidermolysis bullosa may play a role in its known risk for skin cancers, subtypes of the condition have been shown to lead to skin squamous cell carcinoma that is more aggressive than in other conditions involving chronic skin scarring(Fine et al. 2009). The ectodermal dysplasia genes show high coexpression with a few melanoma-altered genes related to cell contact in the epithelium, especially PTK6 (Figure 4-9).

0.8

0.6

0.4

0.2

−0.2

PTK6 Pearson coexpression All genes Ectodermal

Figure 4-9 Coexpression of ectodermal dysplasia genes with PTK6

PTK6 is a recurrently amplified gene in melanoma,. Its coexpression with all genes is compared against its coexpression with the genes associated with the comorbid disease set ectodermal dysplasias including epidermolysis bullosa. Outliers are removed. The two-tailed rank-sum p-value, controlled for number of cancer genes, is 2.0x10-6.

The gene PTK6 is focally amplified in 44% of melanomas and has an identified role in epithelial invasion and mesenchymal transition in prostate and breast cancers(Brauer and Tyner 2010;

Zheng et al. 2013), but the gene has been rarely studied in melanoma. The TCGA melanoma cohort is primarily composed of metastasis samples, but the expression data also includes 103 primary tumors, mostly stage IIC, along with 368 metastases. As changes in cell contact and mesenchymal transition may be related to metastasis state, we compare expression in primary versus metastasis.

We use TCGA barcodes (01 for primary tumor, and 06 or 07 for metastasis), to identify the metastasis and primary samples. We use edgeR to calculate library size factors and estimate dispersion, followed by assessment of differential expression. We find that PTK6 is significantly differentially expressed (adjusted p-value = 3.29x10-28). Then, we examine whether the set of ectodermal dysplasia genes show differential expression, using voom (Law et al. 2014) to transform the data, allowing use of the camera gene set score(D. Wu and Smyth 2012).

Additionally, we use the limma(Smyth 2004) differential expression t-statistic to form a pre- ranked input to GSEA(Subramanian et al. 2005) for gene set differential expression analysis. Of

11 ectodermal dysplasia candidate melanoma genes, nine are significantly downregulated in metastases as compared to primary (gene set differential expression camera p-value = 0.00032,

GSEA p-value = 0, Figure 4-10).

Figure 4-10 GSEA plot of the ectodermal dysplasia candidates

Differential expression is in primary (upregulated) versus metastasis samples.

The other cancers included in our study also have informative genetic and clinical links with

Mendelian disease. Diamond-Blackfan anemia, a blood disorder, is comorbid with the brain neoplasms (ICD10 D61.01, relative risk 95% CI = (9.22 - 28.67)). Indeed, Diamond-Blackfan patients have risk for seven of the cancer ICD-9 code groups, along with other blood and solid cancers(Vlachos et al. 2012). Among Diamond-Blackfan’s causal genes is RPL5, a gene that is significantly deleted in 8% of TCGA glioblastoma and that suppresses MDM2(Dai and Lu 2004)

(Figure 4-11a). MDM2 is recurrently amplified in 15% of TCGA glioblastoma cases. It is an established oncogene that negatively regulates TP53(Manfredi 2010). Like RPL5, other

Diamond-Blackfan genes RPL11 and RPS7 repress MDM2 in response to ribosomal stress(Manfredi 2010). The deletion of RPL5 is mutually exclusive with amplification of MDM2

(p=0.033, Figure 4-11b), supporting the role of RPL5 deletion as an alternative mode of TP53 abrogation. While RPL11 is less frequently deleted, it also has a mutually exclusive pattern with

MDM2 amplification (p=0.042). The role of these ribosomal proteins in glioblastoma appears to be unstudied, making this an exciting novel finding.

a b RPL11 RPS7 RPS7% MDM2% RPL5 RPL5% MDM2 TP53% RPL11% 561 GBM copy number profiles D61.01 Diamond-Blackfan

Figure 4-11 Interaction of Diamond-Blackfan anemia genes with glioblastoma altered genes. a. Summary of genes and their known interactions. b. Summary of copy number changes to MDM2 and the Diamond-Blackfan associated ribosomal proteins known to suppress the action of MDM2. Among the ribosomal genes, RPL5 is recurrently and focally deleted such as to be in the GISTIC2 results, and it shows mutual exclusivity with MDM2 amplification. RPL11 deletion is less frequent but it is also mutually exclusive. RPS7 and RPL11 deletions, together with RPL5 deletions, form a weaker mutually exclusive trend with MDM2 (p = 0.060).

While Diamond-Blackfan anemia is comorbid with many cancers, the cranial development disorder holoprosencephaly is only comorbid with the brain neoplasms (ICD10 Q04.2, relative risk 95% CI = (9.30 - 15.95)). Defects in genes that regulate cranial-specific components of the sonic hedgehog pathway are responsible for the improper embryonic patterning in holoprosencephalies(Taniguchi et al. 2012). This pathway regulates expression of the GLI transcription factors, which have been linked to maintenance of stemness in gliomas(Clement et al. 2007). Subtypes of glioblastoma have been defined on the basis of gene expression patterns, and among these the Classical subtype has a signature including Sonic hedgehog signaling(Verhaak et al., 2010a). Holoprosencephaly genes have weak pathway enrichment similarity with low-grade glioma genes, as well as coexpression with multiple of the low-grade glioma genes, particularly the recurrently copy number altered gene VENTX (p = 0.0092). In the

TCGA lower grade glioma cohort, VENTX lesion occurs more in higher grade tumors, and these lesions are anticorrelated with IDH1 mutation. Mutation of IDH1 is associated with good prognosis and particularly co-occurs in subtype of low grade glioma with either TP53 alteration or 1p19q codeletion(Bourne and Schiff 2010). Comparing the IDH1 mutated against the VENTX mutated samples for patients with both mutation and expression data available, we find strong differential expression of the holoprosencephaly genes TGIF1, SIX3, ZIC2, GLI2. We use the same methods as detailed previously to assess differential expression of the set of genes. As a set, the holoprosencephaly candidate brain neoplasm genes are significantly upregulated in the

VENTX mutated tumors (camera p-value = 0.048, GSEA p-value = 0.031,Figure 4-12). Both

VENTX mutation and activated hedgehog signaling are thus associated with higher grade gliomas.

Changes in regulation of the sonic hedgehog pathway may be an important step in the progression of lower grade glioma, as is known to be true in the Classical subtype of glioblastoma.

Figure 4-12 GSEA plot of holoprosencephaly candidate genes

Differential expression of these genes is compared in the VENTX copy number altered samples (upregulated) versus the IDH1 mutated samples. The hedgehog related genes are upregulated in the VENTX altered samples.

4.3.2 Pan-cancer Mendelian associations Above, we describe a number of processes aberrantly regulated in Mendelian disease and in common cancer. The Blair analysis(Blair et al. 2013) suggested that the unique set of Mendelian diseases comorbid with a complex disease represented a sort of barcode, indicative of the unique set of cellular processes underlying each disease. This hypothesis indeed is reflected in the sets of disorders, and underlying genetic lesions, found in this study.

On the other hand, some Mendelian diseases predispose carriers to many cancer types, while others have no relationship with cancer. In fact, the number of comorbid cancers per Mendelian

disease follows a highly non-random distribution (Figure 4-13).

0.3 expected at random 0.25 observed 0.2

0.15

0.1

0.05

Fraction of Mendelian diseases 0 0 2 4 6 8 10 12 Number of cancer comorbidities per Mendelian Disease

Figure 4-13 The distribution of the number of comorbid cancer diagnosis codes per Mendelian disease

The actual distribution (red bars) includes a large number of Mendelian diseases with no cancer relationship, and a long tail with Mendelian diseases that are comorbid with many cancers. The blue bars represent the expected distribution: about one-third of the pairs of disease have a comorbidity relationship, thus the expected mode of the distribution would have four comorbid cancers per Mendelian disease. The expected distribution is modeled using a binomial.

One interpretation of this pattern is that the genes altered in some Mendelian diseases, such as Li-

Fraumeni syndrome, Rubinstein-Taybi syndrome, and Diamond-Blackfan anemia, are related to pan-cancer processes common to cancer development in many contexts. This interpretation is supported foremost by our finding of statistically significant genetic similarity in comorbid disease pairs. Additionally, we examine four new cancers with available TCGA data but no comorbidity information. If the pan-cancer Mendelian diseases impact core cancer processes, we would expect these to be relevant to these new cancers. We test whether pathways associated with

Mendelian diseases with many (more than five) cancer comorbidities are enriched in the four new cancers. We find that the Mendelian diseases with multiple comorbidities share 20 pathways with the four cancers with no comorbidity information, more than the random expectation (p = 0.051, excluding Mendelian cancer syndromes). In another test of this hypothesis, we assess whether

Mendelian diseases with more cancer comorbidities are associated with genes that have cancer-

related characteristics. We create a set of the 48 genes recurrently altered in more than four of the

19 TCGA tumor types. We call these the multi-cancer mutation genes. Examining FANTOM5 coexpression of the Mendelian disease genes and the multi-cancer mutation genes, and we find a significant correlation with number of cancer comorbidities in the gene's associated Mendelian disease. That is: the more cancers that are comorbid with a Mendelian disease, the higher is the coexpression of a Mendelian disease gene and multi-cancer mutation genes (Spearman correlation p-value = 0.027). These findings suggest that some Mendelian diseases predispose patients to many cancers by genetic alteration affecting pan-cancer processes.

The Mendelian diseases with the most links to cancer indeed impact pathways shared across many cancers, including telomere maintenance, DNA damage response, and mTOR signaling 10

(Figure 4-14, and Supplementary Table 3).

Coexpression metric comorbid 5 Pathway metric no comorbidity BioGRID metric unmeasured HumanNet metric Gene enrichment

0 0 5 10 15 Tuberous Sclerosis

Diamond−Blackfan Anemia

Retinitis Pigmentosa

Li Fraumeni and Related Syndromes

Combined Heart and Skeletal Defects

Hereditary Sensory Neuropathy

Lipoprotein Deficiencies

Chronic Granulomatous Disease

Specified Hamartoses

Neurofibromatosis

Disorders of Urea Cycle Metabolism

Spinocerebellar Ataxia

Hypopituitarism

Inherited Anomalies of the Skin

Severe Combined Immunodeficiency

Congenital Ichthyosis

"Polycystic Kidney, Autosomal Dominant" OV LGG KIRP GBM KIRC KICH STAD BLCA LUAD LAML LUSC THCA PRAD BRCA READ COAD UCEC HNSC SKCM

Figure 4-14 Mendelian diseases with broad cancer links

Those Menndelian diseases that have comorbidity with and genetic similarity to more than 3 cancers are compared to all 19 available TCGA cancers, 15 of which have comorbidity information. These mostly have widespread comorbidity and show genetic similarity (after multiple testing correction) across many cancers. Similarity was calculated here without removing the known germline-associated cancer genes in order to view all associations.

Pan-cancer associations with immunodeficiency syndromes could be due to the compromised immune system, rather than the ability of the tumor to evade immune suppression. However, we find many instances of genetic similarity with cancer, suggesting that the same functions are

frequently somatically altered in tumors. For example, the gene B2M is recurrently mutated or deleted in the TCGA melanoma, lung squamous cell carcinoma, and colon adenocarcinoma. Loss of this gene leads to abolition of the MHC class I complex in tumor cells and has been shown to influence immune escape in some lymphomas(Challa-Malladi et al. 2011). B2M has significant coexpression with the immunodeficiency genes, and CIITA and RFX5, immunodeficiency genes that mainly regulate MHC class II expression, have a secondary role in regulating MHC class I expression(Kobayashi and van den Elsen 2012). Novel pan-cancer associations include the set of lipoprotein deficiencies, defects in widely expressed proteins that lead to imbalance of blood cholesterols. The genes associated with lipoprotein deficiencies also influence inflammation and are enriched in the highly cancer-relevant TGF-β pathway. Cancers, with their elevated rates of proliferation, are thought to have high cholesterol metabolism, and the role of blood cholesterol in tumor progression is a current area of research(Llaverias et al. 2011). The lipoprotein deficiency genes are significantly coexpressed with a number of metabolism related genes that are recurrently mutated in multiple cancers (Supplementary Table 3). These include IDH1, a gene that has been shown to be regulated with cholesterol levels(Shechter et al. 2003) and to be relevant in gliomas and other cancers(Turcan et al. 2012). If pan-cancer Mendelian associations exist, this further supports the hypothesis that comorbidity between Mendelian disease and cancer is due to shared processes disrupted by germline or somatic alterations, respectively.

4.4 Discussion

We have shown that Mendelian diseases that are comorbid with a cancer are likely to involve mutation of genes similar to those that are somatically altered in that cancer. Importantly, this suggests that comorbidity between Mendelian disease and cancer may be due to germline mutations that provide a fertile ground for growth of certain aberrant cells. This novel finding provides new insight into the somatic genetic alterations present in a cancer, presenting them in the context of well-characterized diseases with simpler genetics. While algorithms for classifying 91

genes as preferentially somatically mutated in a cancer are an active area of research, comorbidity can provide an orthogonal line of evidence for involvement of some cellular processes in oncogenesis and pinpoint driver genes among the recurrently mutated genes. Candidate drivers among the Mendelian disease genes include many genes that are less recurrently somatically mutated, but that impact the same pathways. Many of our candidate drivers have a bulk of evidence supporting their role: beyond our findings related to comorbidity and genetic similarity, the candidate genes include some recurrently mutated in cancer, and some with identified roles as drivers in other tumors. Additionally, we have used patterns of co-occurrence of candidate mutations across tumor cohorts to demonstrate a likely role for these genes in the tumors. For less frequently mutated candidate drivers, we have related gene expression with clinical indicators.

Our results are informative of the many processes that are involved in cancer development.

Inactivation of ribosomal protein RPL5, associated with Diamond-Blackfan anemia, has the potential to cause aberrant TP53 degradation in multiple cancers. As cancer is known to involve defects in differentiation(Hanahan and Weinberg 2011), much like a number of Mendelian diseases, a role for the Mendelian disease genes in cancer dedifferentiation and aberrant proliferation is plausible. Other “hallmarks of cancer”, such as invasion or regulation of apoptosis are also represented in the Mendelian diseases. As cancers have many altered processes in common, it is logical that we also find some “pan-cancer” Mendelian diseases with multiple genetic and clinical associations.

In contrast, some germline variants predispose patients to a more narrow range of cancers, which can reveal more specific oncogenic processes. A few Mendelian disorders are comorbid only with brain neoplasms and melanoma. As melanocytes are descended from the neural crest, Mendelian genetic lesions affecting neural development are likely to affect processes in melanocytes,

including proliferation and terminal post-mitotic differentiation. One interesting example is microphthalmos, meaning small eye, a disease phenotype that, in the mouse, gave rise to the name of the melanoma oncogene MITF (microphthalmos transcription factor). In humans, the most common causal genes are closely tied in expression and in function to MITF(Adameyko et al. 2012) (Figure 4-6 ). Some of the microphthalmos genes have been implicated in neural derived tumors(Bunt et al. 2010; C. G. Li and Eccles 2012; Yamamoto, Abe, and Emi 2014), and these may be exciting novel candidates in melanoma. There is a link between some sensineural disorders and pigment anomalies: the phenotype of microphthalmos can also occur to varying degrees in patients with Rubinstein-Taybi syndrome and in patients with Waardenburg syndrome, a pigment and deafness disorder. The idea that disorders comorbid with the same cancer may share pathways with each other is highly intriguing. Waardenburg syndrome (included in ICD10 code group Q79.8), like microphthalmos, shows comorbidity only with melanoma and brain neoplasms. Waardenburg has correlated pathway enrichment to melanoma (p = 5.8x10-4): both diseases are impact melanocyte development and β-catenin signaling pathways. However, the billing code used is not specific enough to have significant enrichment.

In fact, many of the Mendelian diseases with an apparent risk for cancer do not display genetic similarity by our metrics. We chose a limited number of genetic similarity metrics in order to consider different lines of interpretable evidence for functional similarity, but other comparisons of genetic similarity could capture more connections. For example, the blood disorder thalassemia can lead to overloaded blood iron levels(Tanno et al. 2007) which may explain these patients’ risk for a variety of cancers(Torti and Torti 2013); however, this effect is not detected by our current approach. Additionally, a number of factors introduce noise into our source data. These issues include ambiguity of the diagnosis codes; heterogeneity of the Mendelian diseases; insufficient sampling of the mutation spectrum of both Mendelian disease and of cancer.

Our finding of statistically significant association of genetic similarity with comorbidity, despite these factors, is a main discovery of our work. This implies that future large scale studies mining rich data sources such as the eMERGE network(McCarty et al. 2011) will find more genetic and clinical associations. Other future work building on our results includes, foremost, the experimental assessment of the novel candidate driver genes. Drugs that target these cellular processes, perhaps as studied in the Mendelian disease patients, may be applicable for the treatment of the tumors(Brinkman et al. 2006).

5 Data-driven discovery of seasonally linked diseases from an Electronic Health Records system

The Electronic Health Record (EHR) is a rich source of data on patterns of human disease. Health records include free text entries as well as coded terms, such as the diagnosis coding system ICD-

9-CM (International Classification of Diseases, Ninth Revision, Clinical Modification). Finding significant disease comorbidities using ICD-9 codes has had implications for the underlying genetic factors of some diseases (Blair et al. 2013) and has been used to suggest unforeseen causes or consequences of disease (Holmes et al. 2011). In another section of the dissertation, I discuss how coded data from clinical records can be combined with cancer genomic information to better understand cellular processes in cancer. In this section of my dissertation I describe an exploratory method to use ICD-9 data to detect seasonal patterns in human disease. Temporal patterns in human disease often reflect changing environmental factors, as is evident in levels of allergic disease in spring and fall, vector-borne and enteric diseases in summer, and respiratory infectious diseases in winter. Thus, discovering temporal associations can potentially inform us of unconsidered causes of a wide variety of human diseases. As EHRs increasingly compile clinical information from large numbers of patients in a computationally accessible form, they represent a unique opportunity to seek these patterns. When this data is explored with appropriate methods, unbiased discovery of trends in incidence could illuminate a diverse array of health conditions.

Additionally, as ICD-9 is an international standard, a uniform methodology could potentially be applied across EHR data from multiple systems. My goal is to examine properties of temporal patterns as observed in ICD-9 codes from the EHR and to relate the discovered patterns to the biology of disease development. This chapter mostly consists of work that was published in

(Melamed, Khiabanian, and Rabadan 2014).

5.1 Introduction

Seasonal disease incidence is often associated with environmental or behavioral risk factors for these maladies, sometimes providing insight into diseases etiology. For example, seasonal changes in levels of allergens influence prevalence of allergies, while infectious diseases such as flu follow a different seasonal pattern. Many recent examples of temporal clustering of disease diagnoses suggest that discovery of seasonality is an important topic. Kawasaki disease, a childhood vascular inflammation that can result in serious cardiac complication, has been recently characterized by a distinct spatiotemporal distribution of cases (Burns et al. 2005; Rodó et al.

2011). In result, the still-uncharacterized pathogenic agent has been suggested to be a windborne microbe. Anecdotal reports of seasonal and weather related patterns of disease incidence have motivated studies on seasonality of heart failure (Gallerani et al. 2011), depression and anxiety

(Winthorst et al. 2011), varicose vein ulcers (Simka 2010), urinary tract infection (Anderson

1983; Falagas et al. 2009), and even cancer (Lambe, Blomqvist, and Bellocco 2003). While some of these works searched for seasonality using purpose-driven surveys, Upshur (Upshur et al.

2005) used coded administrative data derived from a large EHR system to investigate whether seasonal peaks in incidence were a common feature in a limited set of the most frequent diagnoses. Some of these findings emphasize the behavioral causes of seasonal changes in hospital visits, underlining the importance of attributing the likely biological versus sociological causes of the patterns.

However, no systematic method has been developed to detect these seasonal patterns in an unbiased broad scale. While some studies have searched for temporal patterns in disease diagnosis, these works have been limited in the scope of the diseases examined and in the ability to distinguish multiple types of novel seasonal patterns. The extensive longitudinal data on diagnoses in the EHR is a unique source for finding trends in incidence of disease. However, 96

despite the promise of this data, and the potentially strong statistical power of these large patient cohorts, inherent biases in these data obscure identification of seasonal trends. The most computationally tractable component of the EHR is the ICD-9 code. These coded diagnoses are primarily entered into the record in order to enable insurance billing, and entry is manual. Thus, they are incomplete and the patterns of ICD-9 code entry may suffer a number of biases.

However, studies have assessed the ability of the ICD-9 to recover patients a wide range of diseases, showing that they have a strong predictive value for diseases including skin infection

(Levine et al. 2013), urinary tract infection (Tieder et al. 2011), acute myocardial infarction

(Coloma et al. 2013), and chronic obstructive pulmonary disease (Stein et al. 2012).

Additionally, previous studies have examined the distribution of ICD-9 code entries over time in order to learn characteristics unique to that disease. Temporal patterns in ICD-9 codes have been used to pinpoint the influences in increased burdens to emergency units(Tang et al. 2010), and to discover patterns in outcomes in high-risk surgeries(Finks, Osborne, and Birkmeyer 2011).

Members of our group have compared the well-characterized annual seasonal patterns in influenza diagnoses against the influenza pandemic of 2009. The novel strain of the flu was found to be associated with an unusual temporal distribution of influenza diagnoses(Khiabanian et al.

2010). The winter peak occurrence of viral in infections is well known; in contrast, bacteria may cause more infections in warmer months(Perencevich et al. 2008). Thus, the EHR may contain signals of seasonal incidence of disease, possibly implicating pathogens or other risk factors influencing hospital admissions.

Another advantage of seasonality as a research question is that this repeating pattern is less influenced by the many biases inherent in ICD-9 codings. However, as described below, multiple factors confound identification of periodicity. In the New York-Presbyterian EHR system, an

evident increase in diagnosis rates is obvious over the span of years observed. Additionally, the total number of hospitalizations is itself influenced by season. After observing these characteristics, we were motivated to propose a method (Lomb-Scargle periodograms in detrended data, LSP-detrend) to correct for biases and robustly identify periodic temporal patterns. To look for these patterns, we use Lomb–Scargle periodograms, a least-squares method to detect periodic signal. LSP-detrend sensitively uncovers periodic temporal patterns after applying these corrections to the data, and it assigns significance to the patterns. Subsequently, we perform the first comprehensive survey of seasonality in hospital diagnoses, as reflected in

ICD-9 code incidence. We apply LSP-detrend to a compilation of records from 1.5 million patients, comprising many million ICD-9 code entries. In result, we quantify the seasonal trend in the 2,805 most common diagnosis codes coded over 12 years in the New York-Presbyterian system.

Of these disorders, about 10% are identified as seasonal by LSP-detrend, including many known phenomena. Performing a literature review on the resulting seasonal discoveries, we find that many others of these confirm reported or well-established patterns, including some relatively rare diseases. For example, we recover the seasonal winter increase in Kawasaki disease that has been reported in other USA locations. One interesting novel finding is a bi-annual increase in acute exacerbations of myasthenia gravis (ICD-9 code 358.01), with peak incidence in late winter and late summer. This discovery has possible significance for this disease: acute exacerbation is a serious, possibly life-threatening, complication of myasthenia gravis. Thus, we searched the EHR for clues as to the cause of this seasonal pattern, using ADAMS(Holmes et al. 2011) to identify diagnoses that are comorbid with the exacerbations in myasthenia gravis. We dissect the causes of this seasonal incidence, proposing that factors predisposing patients to this event vary through the year. Although EHR data, and ICD-9 coded records in particular, were not created with the

intention of aggregated use for research, these data can in fact be mined for periodic patterns in incidence of disease, if confounders are properly removed. This work points to the potential of the EHR as a source for unbiased pattern discovery, with implications for understanding human disease.

5.2 Methods

5.2.1 Quantifying incidence of diagnoses We start from a de-identified data set from the New York Presbyterian EHR system, previously used in(Holmes et al. 2011), that includes a patient number, date of visit, and diagnosis code.

After reviewing the total number of cases over the entire period recorded in the EHR, we restrict our analysis to hospital visits happening from 1996, as the number of records drops significantly before this year, until 2009, the last year available. From this data set, we extract all diagnoses with more than 500 unique cases across that time period. Finally, the number of unique patients presenting with the diagnosis per month comprises the input to the LSP-detrend analyses.

5.2.2 Correcting for confounding trends First, we examine the total trend of hospital visits by summing the number of cases of each diagnosis together for each month. As shown in Figure 5-1A, and described in greater detail in

5.3.1, the number of cases increases steadily over time. These larger changes obscure the smaller scale periodic pattern: periodograms measure the change from the mean signal as a function of time. Before trend removal, no seasonal pattern is detected, but after the trend is removed, a seasonal pattern in aggregate hospitalizations is evident (Figure 5-1B). Thus, the first step of

LSP-Detrend creates a smoothed version of this large scale pattern, representing the overall trend.

Subtracting the large scale trend from the data results in a “flattened” version of the diagnosis data, with no large scale trend. The trends differ widely per diagnosis. For each code individually, we calculate the smoothed trend at every month using the kernel density estimation implementation from MATLAB which also estimates the appropriate bandwidth(Bowman and

Azzalini 1997). We remove the months two kernel bandwidths from the beginning and end of the 99

entire time period, as they cannot have reliable density estimates. Then, we subtract out the smoothed estimate from the observations in order to create an incidence data set with no overall trend.

The second step of LSP-Detrend removes the seasonal hospital visit trend, which is also described in 5.3.1. This step makes use of the summed number of total hospital visits, once its large scale trend is removed. For each diagnosis code, we mean scale this total hospital load to match the mean number of cases of the diagnosis. This provides an estimate of number of diagnoses of a disease per month that would occur if that diagnosis was always a fixed proportion of the total hospital load. We subtract this monthly estimate of the seasonal hospital trend from the “flattened” data to remove this overall hospital trend.

5.2.3 Evaluating periodicity The method that we chose for assessing periodic signal is the Lomb-Scargle periodogram. This method was first developed for assessment of periodicity when temporal observations are unevenly spaced(Lomb 1976; Scargle 1982). The computed periodogram evaluates the predictive power of each tested frequency. Their work showed that the null distribution of the periodogram for a frequency has follows an exponential, enabling assessment of statistical significance for a given power (Scargle 1982). Using the corrected data described in the previous section, we apply a MATLAB implementation of the Lomb-Scargle method(Press 1992). We discard any significance assigned to periodic signals of with a more than 1.5 year period: these longer periods are less interpretable and as the detrended data only spans a 10 year period, these signals are less well supported by the data. We test 2,805 diseases for periodic patterns of incidence, and then we use the Benjamini-Hochberg procedure (also implemented in MATLAB). We find that a Lomb-

Scargle p-value of < .01 has an expected false discovery rate of less than 15%.

100

5.2.4 Comorbidity analysis One interesting novel result is seasonality in acute exacerbations of myasthenia gravis, as described in 5.3.4. We wish to characterize factors that might influence this seasonal pattern, by looking for comorbid events in the EHR. In previous work from our group, my colleagues developed an algorithm, ADAMS, for identifying comorbid disorders that are specifically associated with a given query disease(Holmes et al. 2011). This method identifies diseases that strongly co-occur with the query disease by comparing the levels of co-occurrence against a given control disease. Finally, the method uses a bootstrap to estimate the false discovery rate.

For this application of ADAMS, we restrict co-occurring diseases to those diagnosed within 60 days before the acute exacerbation event, with a goal of capturing factors likely to have immediate influence on, or to closely reflect, a patient’s state in the lead-up to this complication.

As ADAMS relies on the idea of comparing comorbidities against control diseases, we select control diagnoses that capture aspects of these patients. Thus, we select controls with no likely direct link to myasthenia gravis, but that occur in patient groups of similar age and additionally are frequently diagnosed in this data set. The first control group is patients with influenza (code

487.1), as it is a very common disease that strikes a wide range of age groups in the winter. An additional control group is patients with hip joint pain (719.45): this condition strikes patients with a similar age distribution as myasthenia gravis, and, like the exacerbations of myasthenia gravis, encounters of hip joint pain increase in the summer. The intersection of ADAMS findings as found using each control provides our results, when conditions directly associated with acute exacerbation are removed. The findings are discussed in 5.3.5, and listed in

Supplementary Table 5.

101

5.3 Results

5.3.1 LSP-detrend: finding periodic signal The EHR system at New York-Presbyterian hospital has been in place for three decades, and it contains health information for 1.5 million patients, including both free text and coded entries for diagnosis (ICD-9), procedures, prescription orders, and lab results. We select the time period

1997 to 2009 because the number of entries before 1997 falls off sharply in comparison. As very rare diseases will not have enough data per month to infer a seasonal pattern, we select only diseases with at least 500 cases over the period considered. The data set contains 2,805 diagnoses with more than 500 cases and we obtain diagnosis date and patient identifier for each instance of the diagnosis. The input to our method is then a count of the number of unique patients diagnosed every month.

Two confounding trends emerge when we consider all diagnoses in aggregate. Identifying and removing these trends, as described in 5.2.2, is a major step in identifying periodic signal. First, it is clear that the number of patients visiting the hospital for any reason steadily increases over time

(Figure 5-1A). We remove this trend for each code by subtracting out a smoothed version of the incidence information, a procedure we call de-trending. Upon removing this trend in the aggregate diagnosis data, we are able to identify with strong confidence a seasonal increase in the number of hospital visits in the spring and in the fall(Figure 5-1B). This hospital visit trend is reflected in the monthly frequencies of the most common diagnoses: the more common a diagnosis is, the more its monthly incidence reflects this overall trend (Figure 5-1C). The most common diseases in the hospital include many chronic diseases, such as Unspecified Essential

Hypertension (401.9), Obesity unspecified (278.00), and Osteoporosis unspecified (733.00). The high prevalence of chronic diseases among the diseases with a spring and fall increase provides a clue as to the meaning of this trend. These diseases are unlikely to be the primary cause of most

102

20 Period 0.5 yrs with FAP of 9.0172e−07 15

0 0 5 10 15 20

x 1010 15 Period = 0.5 seasonal10 hospital visits. A more likely explanation is that hospital visits increase seasonally for a number of reasons and these common diseases simply represent a fixed proportion of the overall 5 population. Thus, in a procedure that we term de-totaling, also described in 5.2.2, we remove the

0 total population0 trend5. 10 15 20

x 105 A 2 1.8 Total diagnoses Trend to remove 1.6 De-trended total 1.4

1.2

0.8

0.6 1996 1998 2000 2002 2004 2006 2008 2010 20 Period 0.5 yrs B with FAP of 9.0172e−07 15

10 power 5

0 0 5 10 15 20 Period (years)

C x 1010 15 Period = 0.5

5 Corr. with totaltrend Corr.

0 Number of cases (x 10,000) 0 5 10 15 20

Figure 5-1: Identifying confounding factors in temporal diagnosis x 105 2 A. Aggregated number of diagnoses from 1997 to 2009 (blue) show a strong increasing trend over time, modeled by the red line. When this trend is subtr1.8 acted out, the remaining signal (magenta) shows no overall trend but a seasonal trend. B. De1.6 -trended total diagnoses display 6 month periodicity, as shown by the periodogram. When the years are plotted on top of each other (each colored line represents a 1.4 year, with the bold black line the average), the spring and fall show consistent peaks in diagnosis each year. C. For each diagnosis, the number of cases is compared with the seasonal pattern in incidence. The 1.2 most frequently occurring diagnoses show the 1 most correlation with the overall spring-fall peak incidence; the overall trend causes false detection of periodic signal. Correlation between number of 0.8 cases (x-axis) and correspondence with the overall spring-fall trend (y-axis) is 0.41. 0.6 1996 1998 2000 2002 2004 2006 2008 2010 103

The final step of LSP-Detrend assesses the adjusted data for periodic signal. We use Lomb-

Scargle periodograms, which use the time series of monthly rate of diagnosis as the input. For a range of possible periods, the power of that period, and an associated statistical significance, is calculated. We find the period of greatest power for the uncorrected data, the de-trended data, and the de-trended and de-totaled data.

5.3.2 Major types of periodic signal and known seasonal disease Of the 2,805 codes in our study, 284 have a significant periodic signal that is likely to represent seasonal peaks in incidence. Performing hierarchical clustering of the monthly occurrence of each disease, we look for groups of conditions with similar period and phase of incidence (Figure

5-2). The clustering shows that two main groups comprise most of the diagnoses.

Above Below average average Summer peak peak Summer Disease(clustered) Winter peak Winter peak

Month (ordered by time over 10 years)

Figure 5-2: Pre-processed and row-normalized monthly incidence for 227 codes with periodic signal.

Each row is a disease, and each column a month over 10 years. Thus, boxes in a row represent incidence of that disease for each month, with red signifying elevated incidence and green decreased incidence. Two main clusters stand out: diseases that occur in the summer (top), and those that occur in winter.

104

The two groups split between events that occur in winter (including viral infections and respiratory infections), and those diagnoses that occur in summer (mostly fractures and wounds).

Based on the groupings of codes, seasonal influences appear to arise from a number of sources.

Many of the patterns appear to be associated with seasonally influenced behavior changes.

However, the majority of these have not been previously reported in the literature. This includes the predominance of accidents in the summer. As well, rashes and skin infections like impetigo are likely linked to more skin exposure in the summer, and the same factor seems likely to explain the increased diagnosis of the chest bone malformity pectus excavatum (code 754.81).

Seasonal behavior change also drives the pattern in diagnoses pertaining to child psychiatric disorders, including attention deficit disorder and adjustment disorders, which dip sharply during the summer school break. A well known annual pattern, the increase in births in the summer, has also been suggested to be most attributable to behavior, though other factors may play a role

(Bobak 2001).

Although seasonal changes in behavior explains many of the temporal patterns in diseases rates throughout a year, environmental risk factors clearly vary as well, including allergens, ultraviolet light, and the virulence of pathogens. It is well known that some allergies, influenza, pneumonia, scarlet fever, and complications from these disease have clear peaks in incidence. All of these effects were captured in our data and by LSP-detrend, showing that ICD-9 codes do reflect these patterns, and that our method is sensitive to such signal. The next section focuses on the findings that appear most novel, interesting, and interpretable.

5.3.3 Confirmation of recent reports of seasonal effects

While no previous method has performed a systematic search for seasonal trends in disease incidence, some previous studies have assessed seasonality of individual diseases. These studies

105

are usually inspired by anecdotal observation of a seasonal association of a disease, and these reports use a variety of methods to scientifically assess these hypotheses. such as questionnaires, mining surveillance databases, examining lab results, or performing patient chart review for a pre-selected set of patients. LSP-Detrend, on the other hand, does not require a pre-defined hypothesis, but instead it is able to rank the seasonality for all diagnoses in the hospital, using existing hospital data. While the previous studies have required extensive data collection, we take advantage of an already rich data source, and we show that LSP-Detrend, and the ICD-9 data, is able to reproduce findings reported elsewhere. The diseases that LSP-Detrend assigns strong signal include a number of previous reports of periodic disease incidence. We compare our findings to these studies below.

First, neuropsychiatric diseases provide a particularly interesting subset, as influences in their occurrence are controversial. Some literature supports seasonal changes in occurrence of anxiety and depression (Winthorst et al. 2011). Taking a mechanistic approach, other groups have documented seasonality of key neurotransmitters involved in mood (Lambert et al. 2002;

Molendijk et al. 2012). Our analysis uncovers a strong winter and early spring increase in obsessive-compulsive disorder (300.3), dysthymic disorder (300.4), shown in Figure 5-3A, and other depressive disorders (311). It is difficult to attribute trends in these complex disorders to behavioral versus environmental influences. Thus, we find an interesting contrast in other psychiatric disorders, such as dependent personality disorder and social phobia. These have no seasonal pattern, suggesting that different factors influence these diseases.

106

20 10 Period 0.9899 yrs 8 with FAP of 0.0093922 15 Period 1.0427 yrs 6 10 with FAP of 0.00040641 4 5 2

0 0 0 5 10 15 20 0 2 4 6 8 10 12 14 16

x 106 2 10000 40 Period = 1.0167 8000 1.5 Period = 1.0104 30 10 6000 Period 0.97778 yrs Period 0.50575 yrs 1 with FAP of 8.8833e−09 20 4000 with FAP of 0.0094043 0.5 10 20005

00 0 00 2 4 5 6 8 1010 12 14 1516 18 20 2022 0 2 4 6 8 10 12 14 16 0 0 1 2 3 4 5 6 7 de−trend, de−totaled 300.4 Dysthymic disorder de−trend, de−totaled 446.1 Acute febrile mucocutaneous lymph node syndrome (mcls) x 107 cc = 0.12 cc = −0.08 2 500 200015 1.5 Period = 0.52381 400 Period = 1 1500 1 10 300 1000 2000.5 5 500 100 0 0 19982 420006 20028 10 200412 14200616 18200820 22 0 1998 2000 2002 2004 2006 2008 0 1 2 3 4 5 6 7 de−trend, de−totaled 599.0 Urinary tract infection site not specified A 300.4 Dysthymic cc = −0.13 disorder Cde −trend,446.1 de−totaled Acute febrile 358.01 mucocutaneous Myasthenia lymphgravis node with syndrome (acute) exacerbation(mcls) 400 20 cc = −0.08 1600 15 1400 15 300 1200 1010 1000 200 800 55 600 100 1998 2000 2002 2004 2006 2008 0 J F M A M J J A S O N D J J 2004F M2005A M2006J J2007A S2008 O 2009N D J

B 599.0 Urinary tract infection site not specified D 1300 358.01 Myasthenia gravis with (acute) exacerbation 15 1200 10 1100 5 1000

900 0 J F M A M J J A S O N D J J F M A M J J A S O N D J

Figure 5-3: Selected diseases with periodic signal.

For four of the diseases discussed, the monthly incidence for all years is plotted together in order to view consistencies in the seasonal trend across different years. Each colored line again represents a year, with the bold black line the average across the years.

Other periodically increasing diseases are likely linked to seasonally increased environmental risk. Recently, reports have asserted that bacterial infections are more frequent in warmer weather

(Perencevich et al. 2008), that bacterial bloodstream infections peak in summer (Eber et al. 2011), and that there is a strong seasonal significant effect on bacteria virulence (Frankel et al. 2012).

Our data strongly support the hypothesis that bacterial infections are higher in the summer. We detect strong summer periodic signal in urinary tract infection (code 599.0), shown in Figure

5-3B, and its complications of pyelonephritis (590.10 and 590.80), and hematuria (599.7). This corroborates results from (Anderson 1983; Falagas et al. 2009). We also detect increased rates of cellulitis and abscess in the summer months. Other groups have found increased incidence of soft tissue infections in the summer (X. Wang et al. 2013). Finally, there is a strong late summer peak incidence of vascular device inflammation and infection (code 996.62), which may be due to the same influences.

107

Finally, as one inspiration for this work is to discover novel pathogenic effects contributing to disease incidence, we are particularly interested in the identified annual winter peak in Acute febrile mucocutaneous lymph node syndrome (mcls) (code 446.1), also known as Kawasaki disease. The winter peak in incidence (Figure 5-3C) is strong and the finding is consistent with previous USA reports(Rodó et al. 2011). Although this disease is rather infrequent in our cohort,

LSP-detrend confidently identifies the pattern.

Thus, the results contribute knowledge of a range of human diseases, and many of our findings are buttressed by previous reports investigating specific hypotheses about disease incidence.

5.3.4 Novel findings: acute exacerbations of myasthenia gravis One of our findings stands out for further analysis. Myasthenia gravis with acute exacerbation, code 358.01, was particularly interesting because it is a well-defined diagnosis, the condition is acute, requiring immediate attention, and the seasonal incidence is previously entirely unreported.

Although this is a rare condition, the seasonal trend is strongly visible in Figure 5-3D, with peak incidences in late winter and in late summer.

Myasthenia gravis is an autoimmune disease characterized by presence of antibodies targeting elements of the nerve to muscle junction. The result is a decay of this neuromuscular junction, resulting in blocked neural signals to the muscle and subsequent muscle weakness (Querol and

Illa 2013). Subtypes include patients with antibodies targeting the acetylcholine receptor, and against the muscle specific kinase receptor, with variable phenotype, including treatment response, depending on the category of autoimmune antibody. Patients, usually older middle- aged people or young women, often primarily present with eyelid weakness (ptosis) or difficulty swallowing (dysphagia) or other signs of weakness. Suggested underlying causes include abnormalities of the thymus, such as thymoma, certain drugs, as well as a genetic component.

108

Although immunosuppressive drugs or thymectomy can mitigate the weakness and allow a high quality of life for most subtypes of patients, the condition is incurable and is characterized by occasional acute episodes. Any infection can cause increased immune activity which can worsen an autoimmune condition, and the role of viral infections in particular have been investigated

(Cufi et al. 2013). Other risk factors include stress, many drugs, and temperature, which directly reduces the efficiency of the neuromuscular junction. In acute exacerbation of myasthenia gravis, a patient experiences greater weakness, sometimes to the point where normal respiratory activity is threatened, resulting in respiratory failure. Thus, patients with this diagnosis have a high likelihood of seeking prompt treatment. The chronic condition, with a different administrative code, shows no seasonal pattern.

5.3.5 Dissecting causes of seasonality of acute exacerbations by finding comorbid diseases As this seasonal pattern has no obvious cause, we use the EHR data to search for common factors among patients with exacerbation. Concurrent diagnoses, prescriptions, or procedures could provide insight into cause of the seasonal pattern. Described previously (Holmes et al. 2011),

ADAMS, Application for Discovering Disease Associations using Multiple Sources, has been used with the same clinical data set to find comorbid diagnoses for rare diseases. We use the method to search for comorbid events specific to the months preceding the exacerbation, as described in 5.2.4, and the results are shown in Supplementary Table 5. When conditions that can be assumed to be a direct result of the myasthenia gravis are disregarded, the most associated diseases are urinary tract infection (code 599.0), carpal tunnel syndrome (code 354.0), unspecified essential hypertension (code 401.9) and esophageal reflux (code 530.81). We find it particularly interesting that both UTI and carpal tunnel are seasonally linked, although the explanation for the pattern in carpal tunnel is unclear. It is possible that treatment for a seasonally linked disease exacerbates the myasthenia gravis, as this condition is worsened by many commonly used drugs. 109

20 Period 0.5 yrs 5.3.6 Comparisonwith FAP of 9.0172e−07 between hospital systems 15 Because ICD-9 is a standard, we finally sought to reproduce our seasonal findings in the Stanford 10

EHR system.5 The Stanford data covers a later time period than the Columbia data did, comes

0 from a geographic0 5area with 10a more moderate15 climate,20 and contains a patient group that is not

x 1010 likely to15 have strong overlap with the Columbia data set. We find that this system has a much Period = 0.5 weaker10 seasonal pattern of overall hospitalizations. While the seasonal pattern is correlated to that observed5 in Columbia,, the variance between years was much higher (Figure 5-4).

0 0 5 10 15 20 COLUMBIA&

x 105 2

1.8 Total&diagnoses& Trend&to&remove& 1.6 De1trended&total& 1.4

1.2

0.8

0.6 1996 1998 2000 2002 2004 2006 2008 2010

STANFORD&

Figure 5-4: Overall seasonality of hospitalization in Columbia and Stanford

On the left, the data across years is plotted, showing the de-trended total hospitalizations in pink. On the right, each year is plotted over the others, with the mean across the years plotted as the thicker black line.

Approximately 25% of the seasonal ICD-9 codes from the Columbia system have similar patterns of significant seasonality in the Stanford system, and an additional number of similar diseases have similar seasonal patterns between the two hospitals. Of note, clear coding differences exist between the hospitals, confounding a full comparison of the patterns. For example, where the

Columbia system has a seasonal increase in urinary tract infection, the Stanford system shows a

110

similar increase in acute cystitis. A number of diseases, including myasthenia gravis with acute exacerbation, have too few patients in the Stanford system for the LSP-Detrend analysis. Other similarities between the two sets of results include: a springtime increase in carpal tunnel and ulnar nerve lesion in both sets; a summer increase in accidents and birth related events; and an increase in complications of procedures in the summer. We conclude that there is some reproducibility, but that it is difficult to confidently identify patterns in rare diseases, and that

ICD-9 can limit identification of patient cohorts.

5.4 Discussion

Applying periodograms to administrative code data from EHR, we are able to identify a number of periodic patterns in an unbiased fashion. Importantly, we demonstrate that examining the data for confounding trends is an essential step for this research question. When the proper corrections are made, the results confirm that LSP-detrend is sensitive to expected seasonal variation, and the method also provides support for recent findings of seasonal distributions of disease. Most significantly among these is a pervasive pattern of increased incidence of bacterial infections in the warmer season, including urinary tract infection, cellulitis and abscess, as well as infection and inflammation of vascular implant. Although some community-acquired bacterial infections in fact are more frequent in the wintertime, it is possible that a distinct subset of bacteria, relying less on community transmission, show more virulence in the warmer weather.

The finding that myasthenia gravis exacerbation has a periodic increase in incidence will be of interest both to clinicians providing care to the patients as well as to immunologists seeking to understand the conditions in which the autoimmune disease is worsened. The results of the comorbidity analysis show that urinary tract infection in particular, as a strong covariate with the condition and as a seasonally linked disease, may have a role to play in exacerbation. It is known

111

that some antibiotics, including those for treatment of urinary tract infection, can worsen the condition. However, as no single correlating factor explains the seasonal pattern clearly, this may be an interesting avenue for further research. There may be an underlying infectious agent causing the immune system flare.

Although ICD-9 codes are primarily recorded for administrative purposes, they have a number of advantages for use in research. Coded data provides a clear categorization of patients and thus this data is a suitable starting point for well-developed computational methods, as compared to other types of EHR information that require inference of disease state. The ICD-9 represent a wide variety of disease, and, importantly, they are an international standard. Thus, a method developed with our system in New York can easily be applied to the myriad other large EHR datasets across the world. In the analysis performed at Stanford, many highly reproducible findings with clinical significance were uncovered. This implies that despite the limitations of

ICD-9 codes, code-based EHR studies offer a promising avenue of research.

EHRs are an increasingly rich source of information. In the future, projects such as the eMERGE network promise the integration of this phenotypic data with genotypic information (Gottesman et al. 2013). With the advent of these resources, patients who display increased encounters for a disease could be interrogated for genotypic markers, allowing us to find new mechanisms of disease, as has been previously investigated in psychiatric disorders. The appropriate methods to analyze this complex source of information in an unbiased fashion holds great promise for human disease.

112

6 CONCLUSION

The approaches I have developed in my PhD studies consider that cancer is the result of a combination of aberrant processes, and signatures of these underlying processes can be found within large collections of genomics and clinical data. Patterns in incidence of cancer are the best source of existing data on cancer biology, yet most methods that mine these data sources have only considered certain types of patterns as interesting. These include signals such as: recurrent selection of mutation(Lawrence et al. 2013; Mermel et al. 2011), mutual exclusivity of mutation(Ciriello et al. 2011; Vandin, Upfal, and Raphael 2012), regulatory relationships(Akavia et al. 2010; Margolin et al. 2006); and projection of genomic data onto annotated gene functional relationships (Sedgewick et al. 2013; Mark D M Leiserson et al. 2014; Hofree et al. 2013).

A unifying theme of my dissertation is finding new ways to integrate the overwhelming variety and quantity of cancer data available. My work has been aimed at expanding the horizon of meaningful patterns in cancer data. In my projects using the total correlation score to find sets of genetic alterations with a related pattern of occurrence, I develop a highly general approach to looking for non-random modules of mutated genes in this data. First, in 3.1, I use total correlation to discover modules in brain neoplasms, an approach that I show can indicate underlying distinctions in the biology of tumor subtypes. Then, in 3.2, I expand on this premise by developing a method that does not use recurrence at all, but instead can identify cancer drivers only by their strong related pattern of alteration with other genes. While cancer is caused by heterogeneous patterns of somatic alterations, Mendelian diseases stand in sharp contrast: these are caused by quite homogeneous and highly penetrant germline variants. But germline mutations that predispose patients to develop particular cancers may, like somatic mutations, represent cellular processes contributing to cancer growth. In chapter 4 I show that comorbidity uncovered

113

from clinical records can be related to processes shared between Mendelian variants and somatic mutations in the comorbid cancer. This finding adds another dimension to TCGA data.

The novel results I have described in this dissertation are also unified by their potential to not only identify key drivers among mutated genes, but to provide insight into the roles of candidate drivers in cancer progression. The results all place genes in the context of related alterations. The total correlation projects mine this large high-quality genomics data for combinatorial patterns of mutations. For example, using GAMToC, I discover a strong connection between TP53 mutations and deletions in the BRSK2 locus on chromosome 11, co-occurring mutations that have not been previously highlighted in literature. Adding Mendelian disease comorbidity allows us to compare the action of disease genes in the context of multiple diseases. In the results from comorbidity and genetic similarity analysis, one interesting gene is PTK6, a recurrently amplified gene in the melanoma cohort that may influence mesenchymal transition in epithelial cells. Among the recurrently altered melanoma genes, this gene stands out for its connection to genes associated with the melanoma-comorbid Mendelian disease epidermolysis bullosa. The epidermolysis bullosa variants impact dermal adhesion and maintenance of basement membrane. The specific phenotype of the Mendelian disease therefore further suggested that in melanoma, PTK6 and the epidermolysis bullosa genes might play a role in invasion and metastasis. This hypothesis was supported in subsequent analysis.

My dissertation work has focused mainly on genetic mutations, specifically copy number alterations and coding sequence mutations. However, many other dimensions are available in cancer genomics data: gene and microRNA expression, methylation, fusion patterns, reverse- phase protein array expression. I have already shown that gene expression patterns of sets of

Mendelian disease genes demonstrate informative signal in their comorbid cancers. This approach

114

used gene expression, and clinical information, as a validation of our candidate Mendelian-related cancer drivers. But, an important improvement on the Mendelian disease project could use this data to predict more candidates. It would also be interesting to apply this idea in the context of the GAMToC projects: sets of co-expressed genes could be identified, and then the combination of sets of gene expression could be assessed using our entropy-based score.

Results from both of my main dissertation projects have provoked further questions that I would like to pursue in my future work. Related genetic mutation events, such as those detected by

GAMToC, are signifiers of convergence in the evolution of cancers across patients. To return to the example of the cell cycle genes mutated in glioblastoma, different subtypes of this disease appear to have specific associations with either CDK4, RB1, or CDKN2A alterations. These genes are all tightly functionally connected, and they have a mutually exclusive pattern of occurrence across the glioblastoma data. While some studies have claimed that this pattern is evidence that the mutations represent alternative equal effects, our result places each of these mutually exclusive alterations in context of other associated mutations. Far from being redundant events, these mutations appear to have specific ramifications affecting cancer progression in different mutational contexts. This is known to be true for changes to the CDKN2A locus, which also influence the TP53 pathway. This is a hint that subtypes of cancer represent combinations of cancer sub-programs, perturbing specific biological processes. Compelling the cellular machinery into a continuous process of division is a shared characteristic that most tumors evolve. But how this function is acquired varies widely. Cancers could acquire this function as a result of germline variants, in the case of Mendelian disease patients with increased cancer risk.

In the case of cancer subtypes, subtype-specific genetic alterations could endow tumors with the needed trait. But what is clear is that the cell cycle trait is “conserved” across tumor cohorts.

115

Most current approaches (Hoadley et al. 2014; Mo et al. 2013; Akbani et al. 2014) searching for signs of convergent evolution across tumors look for hidden classes of tumors. A common approach, achieved through widely varying means, is to cluster tumors regardless of tissue type.

In contrast, very few methods, such as PARADIGM (Sedgewick et al. 2013), quantify the processes in each tumor individually. As each tumor evolves through random mutation followed by selection, treating tumors as members of unified subclasses is a limited approach that can miss the interesting exceptions to this rule. But because tumors clearly have much in common, treating each cancer as unrelated to the others also neglects a large source of information.

Therefore, an approach that I hope to pursue would involve treating individual tumors as a combination of events that are similar to those present across multiple tumors. In this manner, we can discover new shared events driving cancer development, and we can understand how these events impact an individual tumor, with potential therapeutic implications. For example, in melanoma the “Mendelian code” of comorbidities indicates pathways such as melanocyte differentiation and dermal adhesion, reflecting the cell of origin and its surrounding environment.

Growing brain neoplasms face a different set of challenges. As the brain cells they arise from have low inherent replicative potential, alterations related to telomerase activity may be essential to these cancers. Mutations impacting telomere maintenance in brain neoplasms are widespread, but occur in diverse, and usually mutually exclusive, fashion including ATRX coding mutations,

TERT promoter mutations, or the alternative lengthening of telomeres mechanism(Remke et al.

2013). We can harness patterns across tumor cohorts to better understand the mechanisms underlying development of each individual tumor. In this way, the results from my dissertation work can be connected to the goals of precision medicine.

More broadly, patterns in cancer data will only become of increasing importance as the amount and type of cancer data grows. New sources of data such as the eMERGE consortium

116

(Gottesman et al. 2013) have enabled interrogation of the connections between genotypes influencing multiple diseases, an approach that has been termed phenome-wide association study(Denny et al. 2013). Eventually, we will have a wider array of such genotypic and phenotypic information for cancer patients and for the population at large. We will soon be able to integrate a cancer patient’s somatic mutations and clinical trajectory with possible germline influences and environmental factors. At the other end of the spectrum, recent studies have sequenced subclonal populations of tumors at the single cell level. Patterns in subclonal evolution of tumors would be expected to follow many of the same principles as are found at the population level. Convergence in this evolutionary process can be identified to find cancer drivers, much as we did with the GAMToC projects. Additionally, combinations of co-occurring subclonal populations can help us understand the mutation profile observed in bulk tumor data. Cancer genomics, and methods to find patterns impacting tumor growth, has potential to decode the complex series of events leading to cancer, telling us about the biology of tumors and thus the treatments that will provide greatest impact.

117

7 Supplementary Tables Supplementary Table 1: Significantly mutated genes in the melanoma cohort, and their mutations across the tumors pat tumor(Var(Depthtumor(Pos((Depthnormal(Var(Depthnormal(Pos((Depthchr:pos3pos r/f genes codons AAs patient36 24 51 0 32 chr7:1404531363140453136A/T BRAF GTG2GAG V600E patient073057 18 50 0 23 chr7:1404531363140453136A/T BRAF GTG2GAG V600E patient073058 9 24 0 34 chr7:1404531363140453136A/T BRAF GTG2GAG V600E patient073232 28 48 0 30 chr7:1404531363140453136A/T BRAF GTG2GAG V600E patient103104 27 53 0 29 chr7:1404531363140453136A/T BRAF GTG2GAG V600E patient103276 42 70 0 29 chr7:1404531363140453136A/T BRAF GTG2GAG V600E patient145 12 46 0 90 chr2:1187328043118732804G/A CCDC93 GCT2GTT A237V patient16 10 35 0 55 chr2:1187159843118715984G/A CCDC93 TCC2TTC S321F patient073232 25 46 0 62 chr1:2068217303206821730G/A DYRK3 GGC2GAC,GGC2GACG396D,G376D patient103104 15 43 0 38 chr1:2068210603206821060C/T DYRK3 CCA1TCA,CCA1TCAP173S,P153S patient073057 12 38 0 54 chr4:1532473453153247345C/T FBXW7 TGG2TAG,TGG2TAG,TGG2TAGW368*,W486*,W406* patient073058 8 25 0 47 chr4:1532473453153247345C/T FBXW7 TGG2TAG,TGG2TAG,TGG2TAGW368*,W486*,W406* patient16 17 41 0 73 chr12:784288137842881G/A GDF3 CAT1TAT H230Y patient36 12 42 0 27 chr12:784311537843115C/T GDF3 GAG1AAG E152K patient16 21 51 1 140 chr12:14798222314798222G/A GUCY2C CCT1TCT P580S patient36 13 28 0 35 chr12:14829863314829863C/T GUCY2C ATG3ATA M291I patient103276 14 37 0 26 chr12:14809526314809526G/A GUCY2C CGT1TGT R464C patient16 37 68 0 121 chr2:1029553453102955345C/T IL1RL1 CCT2CTT,CCT2CTTP37L,P37L patient103276 10 38 0 37 chr2:1029655423102965542G/A IL1RL1 GGA2GAA G374E patient073057 7 23 0 30 chr5:35876230335876230G/A IL7R GGA2GAA G341E patient073058 6 24 0 18 chr5:35876230335876230G/A IL7R GGA2GAA G341E patient103104 16 47 0 32 chr17:51900681351900681C/T KIF2B TCC2TTC S96F patient103276 10 19 0 20 chr17:51900950351900950G/A KIF2B GAA1AAA E186K patient145 20 71 1 78 chr6:63990011363990011C/T LGSN CGA2CAA R482Q patient16 20 41 0 151 chr6:63990305363990305C/A LGSN TGG2TTG W384L patient16 7 22 0 81 chr6:63991041363991041T/C LGSN AGA1GGA R139G patient16 44 97 0 188 chr4:1642720423164272042C/T NPY5R TCA2TTA S206L patient103104 29 83 1 90 chr4:1642719573164271957C/T NPY5R CAC1TAC H178Y patient145 17 52 0 47 chr8:32621309332621309G/A NRG1 GAT1AAT,GAT1AAT,GAT1AATD443N,D435N,D438N patient103104 26 82 0 62 chr8:32453481332453481G/A NRG1 CGA2CAA,CGA2CAA,CGA2CAA,CGA2CAA,CGA2CAAR294Q,R79Q,R79Q,R79Q,R79Q patient36 7 21 0 13 chr11:479109034791090G/A OR51F1 CCT1TCT P20S patient103104 11 20 0 46 chr11:479061634790616G/A OR51F1 CAC1TAC H178Y patient145 24 66 0 85 chr7:1424585443142458544C/T PRSS1 TCA2TTA S60L patient16 42 167 0 260 chr7:1424584203142458420G/A PRSS1 GAT1AAT D19N patient073058 26 95 0 98 chr9:33796691333796693GAG/3 PRSS3 del E(88388)3 patient073232 16 74 6 100 chr9:337979283337979283/C PRSS3 AGG2+,AGG2+R158+,R101+ patient145 24 52 1 75 chr13:32376340332376340C/T RXFP2 CCA2CTA P688L patient103276 7 32 0 33 chr13:32365959332365959C/T RXFP2 CGA1TGA R388* patient16 14 39 0 73 chr2:2188703218870C/T SH3YL1 GAA1AAA E228K patient073057 11 24 0 41 chr2:2310823231082C/T SH3YL1 GAA1AAA E119K patient073058 6 27 0 36 chr2:2310823231082C/T SH3YL1 GAA1AAA E119K patient36 17 40 0 23 chr19:52002863352002863T/G SIGLEC12 ACG1CCG T306P patient073057 7 15 0 18 chr19:51995082351995082C/T SIGLEC12 GGA2GAA G534E patient16 12 24 0 48 chr6:13588577313588577C/T SIRT5 CGA1TGA,CGA1TGAR44*,R44* patient073058 39 104 0 149 chr6:13612078313612078A/G SIRT5 GAA2GGA E305G patient36 50 175 1 148 chr3:39432984339432984C/T SLC25A38 TCT2TTT S110F patient073232 30 39 0 105 chr3:39433013339433013C/T SLC25A38 CCC1TCC P120S patient145 13 44 0 41 chr20:42694523342694523G/A TOX2 GGC1AGC,GGC1AGC,GGC1AGCG336S,G360S,G378S patient145 13 44 0 40 chr20:42694524342694524G/A TOX2 GGC2GAC,GGC2GAC,GGC2GACG336D,G360D,G378D patient103104 13 27 0 33 chr20:42635250342635250C/T TOX2 CTC1TTC,CTC1TTC,CTC1TTCL35F,L86F,L77F patient103276 16 31 0 16 chr20:42682943342682943C/T TOX2 TCG2TTG,TCG2TTG,TCG2TTGS177L,S228L,S219L patient073057 14 45 0 48 chr2:2345454183234545418G/A UGT1A10 GAA1AAA E84K patient073058 11 45 0 41 chr2:2345454183234545418G/A UGT1A10 GAA1AAA E84K patient145 7 18 0 38 chr3:1672489563167248956T/A WDR49 AAT2ATT N370I patient16 11 27 0 38 chr3:1672457473167245747G/A WDR49 TCA2TTA S470L

118

Supplementary Table 2: Genes significantly less frequently mutated in the nevus cohort, see 2.2.2.4 melF(279) nevF(32) p mutsig THSD7B 151 0 1.09E=10 0.1663559 XIRP2 173 2 4.46E=10 0.04012697 USH2A 147 1 7.76E=09 0.04012697 PTPRT 113 0 1.96E=07 0.09801548 MYH1 99 0 2.09E=06 0.1379674 DSP 90 0 8.81E=06 0.1761339 TPTE 88 0 1.20E=05 0.1529721 SYNE1 114 2 3.08E=05 0.1158377 KCNB2 79 0 4.69E=05 0.1962425 COL3A1 72 0 0.00013 0.00882006 PCDH18 71 0 0.00015 0.03796533 BCLAF1 62 0 0.00052875 0.0220468 DSG3 62 0 0.00052875 0.03248277 PDE4DIP 62 0 0.00052875 0.167958 SNCAIP 61 0 0.00060636 0.1663559 ROS1 77 1 0.00072576 0.1954481 TCHHL1 54 0 0.00155643 0.01057947 ACSM2B 53 0 0.0017768 0.03575059 TP63 53 0 0.0017768 0.03657042 ANO4 51 0 0.00231181 0.03657042 NBPF1 44 0 0.00571202 0.01783554 NRK 44 0 0.00571202 0.00548988 THEMIS 43 0 0.00648653 0.09676519 ARID2 42 0 0.00736234 1.24E=07 STK31 42 0 0.00736234 0.1947338 RUNX1T1 41 0 0.00835224 0.167958 TP53 41 0 0.00835224 4.57E=13 NF1 40 0 0.00947053 2.53E=10 NBEAL1 39 0 0.01073327 0.1407236 PDE1A 39 0 0.01073327 0.00039644 ADAM30 38 0 0.01215843 0.1407236 KEL 38 0 0.01215843 0.01783554 SELP 37 0 0.01376616 0.06169981 POTEG 36 0 0.01557899 0.169217 SLC38A4 35 0 0.01762213 0.00190358 MPP7 34 0 0.0199238 0.07358765 NRAS 85 4 0.02196215 4.57E=13 OR51S1 33 0 0.02251551 0.02791723 CDKN2A 31 0 0.02871411 4.57E=13 EPHA3 31 0 0.02871411 0.1407236 MLL 43 1 0.03979048 0.1262187 NFASC 27 0 0.04644532 0.1179532

119

Supplementary Table 3: Pairs of comorbid and genetically similar Mendelian disease and cancer, related to 4.3. Columns described below: gene_enriched: The corrected significance of the number of genes in common.

geneIntersection: This shows the common genes, even if not statistically enriched

pathway_correlation and pathway: "pathway_correlation" shows the Spearman p-value for the pathways correlation for the pair of diseases, after correction for 427 tests. If this is less than .1, and if there are any shared pathways (significantly enriched in the cancer and impacted by the Mendelian disease), the pathways are shown in the "pathways" column. The format for each pathway shared is: Mendelian_gene_1_in_pathway, Mendelian_gene_2_in_pathway -> Pathway_name (Cancer_gene_1_in_pathway, Cancer_gene_2_in_pathway);... coex_CG and coexpression: "coex_CG" shows the best coexpression score for the pair, corrected across all coexpression results. If this corrected values is less than .1, all cancer genes showing coexpression with the Mendelian disease genes are shown in the "coexpression" column. Each significant cancer gene, is displayed along with any of the Mendelian genes with significant correlation with the cancer gene (rho > .2 for p < .05/number of gene pairs tested). Format: Mendelian_gene_coexpressed_1 -> Cancer_gene_1(Cancer_gene_1 corrected ranksum p-value). Some cancer genes have no Mendelian gene coexprssed at rho > .2, but the set of Mendelian genes still have significantly elevated coexpression. Format: -> Cancer_gene_2(Cancer_gene_2 corrected ranksum p-value). humannet_set and humannet: "humannet_set" shows the corrected p-value for the connections between the disease pair's genes is shown, as described in methods. "humannet" shows all connections. Format: Mendelian_gene_1 -> Cancer_gene_1, Cancer_gene_2; Mendelian_gene_2 -> Cancer_gene_3 …

biogrid_set and biogrid: identical to the humannet columns, but performed on the BioGRID network

120

MD C gene_enrichmentgeneIntersection pathway_correlationpathway coex_CG coexpression humannet_sethumannet biogrid_set biogrid Chronic(Granulomatous(Disease BLCA 1 1 0.078547322 NCF4(@>(BCL2L1(1.69e@02);NCF2,NCF4,CYBB,CYBA(@>(DIAPH2(1.47e@02);NCF2,NCF4,CYBB,CYBA(@>(AHR(1.24e@02);NCF2,NCF4,CYBB,CYBA(@>(PTEN(1.47e@02);NCF4,CYBA(@>(LRP5L(2.06e@02);NCF2,NCF4,CYBB,CYBA(@>(UBXN11(1.91e@02);NCF2,NCF4,CYBA(@>(ZNF586(4.38e@02);NCF2,NCF4,CYBB,CYBA(@>(SH3BGRL3(1.68e@02);NCF2,NCF4,CYBB,CYBA(@>(LPAR6(2.13e@02);CYBA(@>(TMEM80(2.68e@02);NCF2,NCF4,CYBB,CYBA(@>(KDM6A(1.51e@02);NCF2,NCF4,CYBB(@>(TGFBR1(1.43e@02);NCF2,NCF4,CYBB,CYBA(@>(IRF7(2.53e@02);NCF4,CYBB,CYBA(@>(CPM(1.31e@02);NCF4,CYBB,CYBA(@>(CD52(1.56e@02);NCF2,NCF4,CYBB(@>(RBM5(2.82e@02);NCF2,NCF4,CYBB,CYBA(@>(SPCS3(4.68e@02);NCF2,NCF4,CYBB,CYBA(@>(ARL8A(1.69e@02);NCF2,NCF4,CYBB,CYBA(@>(ARHGAP30(1.41e@02);NCF4(@>(PTPN7(3.47e@02);NCF2,NCF4,CYBB,CYBA(@>(CREBBP(1.41e@02);NCF2,NCF4(@>(KAT6A(3.73e@02)1 1 Congenital(Ichthyosis BLCA 1 1 0.001525602 ALDH3A2,ABCA12(@>(FOXQ1(2.31e@02);CSTA,NIPAL4,LIPN(@>(AHR(2.31e@02);SPINK5,CSTA,KRT2,ABCA12,TGM1(@>(PVRL4(6.32e@04);ALOX12B,SPINK5,CSTA,ABCA12,TGM1(@>(LCE3D(5.41e@05);ALOX12B,SPINK5,CSTA,ABCA12,TGM1(@>(LCE3E(5.42e@05);ALOX12B,CSTA,ABCA12(@>(LCE3C(4.15e@04);ABCA12,TGM1(@>(EGFR(2.21e@02)1 1 Polycystic(Kidney,(Autosomal(Dominant BLCA 1 0.096035489 TSC2(@>(Direct(p53(effectors(BCL2L1,TP53,EGFR,PTEN,AFP,CREBBP)0.77024073 1 1 Diamond@Blackfan(Anemia BLCA 1 0.383933529 0.006200982 RPS26(@>(CDKN2A(8.89e@03);(@>(AMMECR1(4.13e@03);RPS19,RPL35A(@>(PABPC4(5.28e@04);RPS26,RPS19,RPS7(@>(TP53(1.79e@03);RPS26,RPS7(@>(UBE2T(4.93e@03);RPS26,RPS7(@>(PFDN2(1.00e@02);RPS26,RPS7(@>(TIMM17A(2.27e@03);RPS26(@>(POU5F1B(3.75e@04)1 1 Inherited(Anomalies(of(the(Skin BLCA 1 1 0.013714977 DKC1(@>(CDKN2A(4.78e@02);NOP10(@>(PABPC4(2.54e@02);KRT6A,NHP2,KRT16(@>(EPS8L2(5.00e@02);WRAP53,DKC1,NHP2(@>(TP53(5.00e@02);KRT6C,KRT6A,KRT1,KRT16(@>(LCE3D(4.68e@02);KRT6C,KRT6A,KRT1,KRT9,KRT16(@>(LCE3E(2.54e@02);KRT6C,KRT6A,KRT9,KRT16(@>(LCE3C(1.00e@03);KRT6A,KRT9,KRT16(@>(EGFR(4.70e@02)1 1 Spinocerebellar(Ataxia BLCA 1 0.151828695 0.000158829 ATXN7,ATM,TBP(@>(ORAOV1(3.31e@06);ATXN7,TDP1,SYNE1,ATM,TTBK2,TBP,ITPR1,PPP2R2B(@>(FHIT(6.28e@03);ZNF592,ATXN7,ATXN2,SYNE1,ATM,TTBK2,TBP,ITPR1,AFG3L2,SETX,PPP2R2B,ATXN1(@>(TNRC6A(1.69e@03);(@>(CHRFAM7A(2.37e@02);JPH3,ZNF592,CACNA1A,ATXN2,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B,ATXN1(@>(CCSER1(2.37e@03);JPH3,CACNA1A,ATXN2,PDYN,SPTBN2,TTBK2,KCNC3,ATXN10,PRKCG,FGF14,SYT14,PPP2R2B(@>(DEAF1(2.43e@03);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(FGFR3(5.62e@03);ZNF592,ATXN7,SYNE1,ATM,TTBK2,ITPR1,SETX(@>(PDE4D(4.83e@02);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(NOVA1(2.41e@03);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(KLHDC9(3.90e@02);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(OPCML(2.99e@03);ZNF592,POLG,ATXN2,TTBK2,TBP(@>(PHRF1(9.19e@03);JPH3,CACNA1A,ATXN2,PDYN,SPTBN2,TTBK2,KC1 1 Hypopituitarism BLCA 1 0.098019822 FGFR1(@>(Syndecan@3@mediated(signaling(events(EGFR,FGFR3);POU1F1(@>(Glucocorticoid(receptor(regulatory(network(TP53,AFP,CREBBP)0.552891743 0.53075862 1 Combined(Heart(and(Skeletal(Defects BLCA 0.53 CREBBP 2.11E@27 CREBBP(@>(inhibition(of(huntingtons(disease(neurodegeneration(by(histone(deacetylase(inhibitors(CREBBP);CREBBP(@>(the(information(processing(pathway(at(the(ifn(beta(enhancer(IRF7,CREBBP);EP300,CREBBP(@>(Direct(p53(effectors(BCL2L1,TP53,EGFR,PTEN,AFP,CREBBP);EP300,CREBBP(@>(p53(pathway(CDKN2A,TP53,CREBBP);EP300(@>(Validated(transcriptional(targets(of(AP1(family(members(Fra1(and(Fra2(CCND1,CDKN2A);EP300,CREBBP(@>(acetylation(and(deacetylation(of(rela(in(nucleus(CREBBP);CREBBP(@>(wnt(signaling(pathway(CCND1,CREBBP);CREBBP(@>(Notch@HLH(transcription(pathway(CREBBP);CREBBP(@>(regulation(of(transcriptional(activity(by(pml(TP53,CREBBP);CREBBP(@>(Presenilin(action(in(Notch(and(Wnt(signaling(CCND1,CREBBP);EP300,CREBBP(@>(Glucocorticoid(receptor(regulatory(network(TP53,AFP,CREBBP);CREBBP(@>(Signaling(events(mediated(by(Stem(cell(factor(receptor((c@Kit)(PTEN,CREBBP);EP300,CREBBP(@>(Signaling(events(mediated(by(HDAC(Class(III(TP53,CREBBP);CREBBP(@>(Signaling(events(mediated(by(TCPTP(EGFR,CREBBP);EP300,CREBBP(@>(FOXM1(tra1 0.02531579 CREBBP(@>(TP53,TP53;EP300(@>(CREBBP0.53565909 Specified(Hamartoses BLCA 0.62 PTEN 0.028031166 PTEN(@>(Direct(p53(effectors(BCL2L1,TP53,EGFR,PTEN,AFP,CREBBP);VHL(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53);PTEN(@>(Negative(regulation(of(the(PI3K/AKT(network(PTEN);PTEN(@>(Signaling(events(mediated(by(Stem(cell(factor(receptor((c@Kit)(PTEN,CREBBP)0.739774885 0 PTEN(@>(EGFR;STK11(@>(TGFBR11 Lipoprotein(Deficiencies BLCA 1 1 0.056770262 APOB,APOA1(@>(AFP(9.01e@03);(@>(UNC93A(1.46e@02);MTTP,APOB,LCAT,SAR1B,APOA1(@>(AFM(6.27e@03);MTTP,APOB,LCAT,SAR1B,APOA1(@>(CDHR5(6.41e@03);MTTP,APOB,LCAT,SAR1B,APOA1(@>(ALB(6.27e@03)1 1 Disorders(of(Urea(Cycle(Metabolism BLCA 1 1 0.078763609 (@>(UNC93A(1.30e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(AFM(1.76e@02);NAGS,ARG1,ASL(@>(CDHR5(1.30e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(ALB(1.76e@02)1 1 Androgen(Insensitivity(Syndrome BRCA 1 0.002753113 AR(@>(Nongenotropic(Androgen(signaling(AKT1,PIK3R1,PIK3CA);AR(@>(FOXA1(transcription(factor(network(FOXA1,CDKN1B,NCOA3);AR(@>(Coregulation(of(Androgen(receptor(activity(CCND1,AKT1,CDKN2A,CASP8);AR(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(CDKN2A,TBL1XR1,CCND1,MED12,CDH1,TERT)1 1 0.2405 AR(@>(NCOA3 Chronic(Granulomatous(Disease BRCA 1 0.407470815 0.076994822 NCF2,NCF4,CYBB,CYBA(@>(RPGR(1.37e@02);NCF2,NCF4,CYBB,CYBA(@>(SELPLG(1.16e@02);NCF2,NCF4,CYBA(@>(GPS2(1.52e@02);NCF2,NCF4,CYBB(@>(VPS9D1(1.65e@02);NCF2,NCF4,CYBB,CYBA(@>(MAP3K1(1.13e@02);NCF4(@>(CCDC18(3.95e@02);NCF2,NCF4,CYBB,CYBA(@>(PTEN(1.24e@02);NCF2,NCF4,CYBB,CYBA(@>(ZNF276(1.19e@02);NCF2,CYBB,CYBA(@>(UBC(1.22e@02);NCF2,NCF4,CYBB,CYBA(@>(KDM5A(2.12e@02);NCF2,NCF4,CYBB,CYBA(@>(NFATC1(1.34e@02);NCF4,CYBA(@>(CDKN1B(4.01e@02);NCF2,NCF4,CYBB,CYBA(@>(HLA@A(1.79e@02);NCF2,NCF4,CYBB,CYBA(@>(NR1H2(1.11e@02);(@>(SPATA12(3.99e@02);NCF2,NCF4,CYBB,CYBA(@>(TICAM1(1.07e@02);NCF2,NCF4,CYBB,CYBA(@>(ITPR1(4.30e@02);NCF2,NCF4,CYBB(@>(MEF2A(3.00e@02);NCF2,NCF4,CYBB,CYBA(@>(RBM23(1.15e@02);NCF2,NCF4,CYBB,CYBA(@>(CTDP1(1.09e@02);CYBA(@>(NCOA3(1.67e@02);NCF2,NCF4,CYBB,CYBA(@>(RUNX1(1.11e@02);NCF2,NCF4,CYBB,CYBA(@>(TCF25(1.79e@02);NCF2,NCF4,CYBB,CYBA(@>(PQLC1(1.12e@02);NCF2,CYBB,CYBA(@>(NEU4(1.24e@02);NCF2,NCF4,CYBB,CYBA(@>(SRXN1(1.12e@02);NCF2,NCF4,CYBB,CYBA(@>(LPAR6(1.44e@02);NCF2,NCF4,CYBB,CYBA(@>(RBM7(1.10e@02);NCF2,NCF4,CYB1 1 Cerebral(Degeneration(Due(to(Generalized(LipidosesBRCA 1 1.93E@05 SMPD1(@>(Ceramide(signaling(pathway(PDGFA,MAP3K1,CASP8,AKT1,RB1,MAP2K4);SMPD1(@>(IL2(signaling(events(mediated(by(PI3K(MYB,TERT,PIK3CA,PIK3R1,AKT1);SMPD1(@>(phospholipids(as(signalling(intermediaries(PDGFA,PIK3CA,PIK3R1,AKT1);SMPD1(@>(ceramide(signaling(pathway(MAP3K1,CASP8,MAP2K4)0.159853943 1 1 Congenital(Ichthyosis BRCA 1 1 0.073521831 ALOX12B,ALOXE3,CSTA,TGM1(@>(PARD6G(3.75e@02);ALOXE3,CSTA,ABCA12,TGM1(@>(CDH1(3.31e@02);SPINK5,CSTA,TGM1(@>(MUC21(2.98e@02);CSTA,NIPAL4,LIPN,ABHD5(@>(UBC(2.59e@02);ALOXE3,CSTA,NIPAL4(@>(PHLDA1(3.75e@02);ALOX12B(@>(KRTAP9@9(9.81e@03);ALOX12B(@>(KRTAP4@5(9.81e@03);CSTA,NIPAL4,LIPN,ABHD5(@>(TICAM1(3.83e@02);LIPN,ABHD5(@>(ZFP36L1(2.60e@02);CSTA,NIPAL4,LIPN,ABHD5(@>(SRXN1(2.98e@02)1 1 Diamond@Blackfan(Anemia BRCA 1 0.968925499 0.001684618 (@>(MYB(3.65e@02);RPS26(@>(CDKN2A(1.09e@02);RPS26,RPS19,RPS10,RPL11,RPL35A,RPS7(@>(SLC25A5(1.93e@04);(@>(E2F4(4.16e@03);RPS26,RPS24,RPL5,RPS19,RPS10,RPL11,RPL35A,RPS7(@>(RPL13(6.67e@05);RPS26,RPL5,RPS19,RPS10,RPL11,RPL35A,RPS7(@>(RPL18(7.97e@05);(@>(NFE2L3(1.40e@03);RPS26,RPS19,RPS10,RPS7(@>(TCF3(2.93e@03);RPS26,RPS19,RPS7(@>(TP53(1.91e@03);RPL11,RPL35A,RPS7(@>(RBMX(2.97e@02);RPS26,RPS7(@>(FANCA(7.27e@03);RPS26,RPS19,RPS10,RPL35A,RPS7(@>(TXNL4A(3.48e@02);(@>(CPNE7(2.04e@02);(@>(TERT(1.31e@02);RPS26,RPS19,RPS7(@>(HIST1H3B(7.83e@03)1 1 Inherited(Anomalies(of(the(Skin BRCA 1 TERT 6.65E@05 ATP2A2(@>(nfat(and(hypertrophy(of(the(heart((NFATC1,PIK3CA,PIK3R1,AKT1);TERT(@>(IL2(signaling(events(mediated(by(PI3K(MYB,TERT,PIK3CA,PIK3R1,AKT1);TERT(@>(overview(of(telomerase(protein(component(gene(htert(transcriptional(regulation(TERT,TP53);TERT(@>(telomeres(telomerase(cellular(aging(and(immortality(AKT1,TERT,TP53,RB1);KRT1,TERT(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(CDKN2A,TBL1XR1,CCND1,MED12,CDH1,TERT);TERT(@>(role(of(nicotinic(acetylcholine(receptors(in(the(regulation(of(apoptosis(PIK3CA,TERT,PIK3R1,AKT1);TINF2,TERT,DKC1(@>(Regulation(of(Telomerase(CCND1,CDKN1B,TERT,AKT1)0.183228108 1 1 Spinocerebellar(Ataxia BRCA 1 ITPR1 0.097855772 PRKCG(@>(EGFR(Inhibitor(Pathway,(Pharmacodynamics(ERBB2,PIK3CA,MAP3K1,PIK3R1,AKT1);ATM(@>(apoptotic(signaling(in(response(to(dna(damage(AKT1,TP53);ATM(@>(role(of(brca1(brca2(and(atr(in(cancer(susceptibility(TP53,FANCA);CACNA1A(@>(Anti@diabetic(Drug(Potassium(Channel(Inhibitors(Pathway,(Pharmacodynamics(AKT1,HNF1A,PIK3R1,PIK3CA);TBP(@>(Validated(targets(of(C@MYC(transcriptional(repression(ERBB2,CDKN1B,CCND1,ZFP36L1);ATM(@>(rb(tumor(suppressor/checkpoint(signaling(in(response(to(dna(damage(TP53,RB1);ATM(@>(E2F(transcription(factor(network(CDKN1B,CDKN2A,MCL1,RB1,E2F4);CACNA1A(@>(rac1(cell(motility(signaling(pathway(PIK3CA,MAP3K1,PIK3R1);ATM(@>(hypoxia(and(p53(in(the(cardiovascular(system(AKT1,TP53);TBP(@>(Glucocorticoid(receptor(regulatory(network(NFATC1,TP53,GATA3,AKT1);ATM(@>(Regulation(of(Telomerase(CCND1,CDKN1B,TERT,AKT1);PRKCG(@>(Retinoic(acid(receptors@mediated(signaling(AKT1,NCOR2,NCOA3);ATM(@>(Validated(transcriptional(targets(of(deltaNp63(isoforms(CDKN2A,RUNX1,AXL)0.031145236 APTX,SYT14,PPP2R2B(@>(LYRM2(2.24e@02);JPH3,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,PRKCG,FGF14,SYT14,PPP2R2B(@>(CXXC11(7.79e@03);(@>(CHRFAM7A(3.35e@02);ZNF592,ATXN7,TDP1,SYNE1,ATM,TTBK2,TBP,SETX,ATXN1(@>(PIK3CA(2.81e@02);JPH3,ZNF592,ATXN7,ATXN2,TDP1,SYNE1,ATM,TTBK2,TBP,KCNC3,FGF14,AFG3L2,PPP2R2B(@>(MAP3K4(3.34e@03);JPH3,ATXN2,SYNE1,TTBK2,SYT14,PPP2R2B(@>(IGF1R(2.05e@02);JPH3,ZNF592,ATXN2,SYNE1,ATM,SPTBN2,TTBK2,KCNC3,PRKCG,SETX,PPP2R2B,ATXN1(@>(WNK1(7.91e@03);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRKCG,FGF14,SYT14,PPP2R2B(@>(KCNN3(8.91e@03);ZNF592,ATXN7,ATXN2,TDP1,ATM,TTBK2,TBP,NOP56,AFG3L2,SETX,C10orf2(@>(CTCF(3.07e@03);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,PRKCG,FGF14,SYT14,PPP2R2B(@>(ASPHD1(1.05e@02);JPH3,CACNA1A,ATXN2,PDYN,SPTBN2,TTBK2,KCNC3,PRKCG,FGF14,SYT14,C10orf2,PPP2R2B(@>(GPR19(1.66e@02);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,PRKCG,FGF14,SYT14,PPP2R2B(@>(IQSEC3(7.97e@03);JPH3,CACNA1A,ATXN2,PDYN,SPTBN2,TTBK2,KCNC3,ATXN10,PRKCG,FGF14,SYT14,PPP2R2B(@1 1 Hypopituitarism BRCA 1 0.008828178 GH1(@>(regulation(of(eif@4e(and(p70s6(kinase(PTEN,AKT1,PIK3R1,PIK3CA);GLI2(@>(Signaling(events(mediated(by(the(Hedgehog(family(AKT1,PIK3R1,PIK3CA);GH1(@>(mtor(signaling(pathway(AKT1,PIK3CA,PIK3R1,PTEN);BTK(@>(Fc@epsilon(receptor(I(signaling(in(mast(cells(AKT1,MAP3K1,PIK3CA,PIK3R1,MAP2K4);BTK(@>(BCR(signaling(pathway(NFATC1,AKT1,PIK3CA,MAP3K1,PTEN,PIK3R1);GH1(@>(growth(hormone(signaling(pathway(PIK3CA,HNF1A,PIK3R1);POU1F1(@>(Glucocorticoid(receptor(regulatory(network(NFATC1,TP53,GATA3,AKT1);GH1(@>(akt(signaling(pathway(PIK3CA,PIK3R1,AKT1);GH1(@>(trefoil(factors(initiate(mucosal(healing(ERBB2,AKT1,MUC2,PIK3R1,PIK3CA);FGFR1(@>(FGF(signaling(pathway(CDH1,AKT1,FGFR2,PIK3R1,PIK3CA)1 1 1 Combined(Heart(and(Skeletal(Defects BRCA 1 2.37E@10 CREBBP(@>(nfat(and(hypertrophy(of(the(heart((NFATC1,PIK3CA,PIK3R1,AKT1);EP300,CREBBP(@>(IFN@gamma(pathway(AKT1,MAP3K1,PIK3R1,PIK3CA);EP300,CREBBP(@>(FOXA1(transcription(factor(network(FOXA1,CDKN1B,NCOA3);EP300,CREBBP(@>(transcription(regulation(by(methyltransferase(of(carm1(NCOA3,PRKAR1B);EP300(@>(Validated(transcriptional(targets(of(AP1(family(members(Fra1(and(Fra2(CCND1,NFATC1,CDKN2A);EP300(@>(Notch(signaling(pathway(NCOR2,CCND1,NOTCH3,MFAP2,GATA3,NCOR1);EP300(@>(Validated(nuclear(estrogen(receptor(alpha(network(CCND1,NCOA3,NCOR2,NCOR1);EP300(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(CDKN2A,TBL1XR1,CCND1,MED12,CDH1,TERT);EP300,CREBBP(@>(mechanism(of(gene(regulation(by(peroxisome(proliferators(via(ppara(FAT1,PRKAR1B,RB1,NCOR1,NCOR2);EP300(@>(Validated(targets(of(C@MYC(transcriptional(repression(ERBB2,CDKN1B,CCND1,ZFP36L1);CREBBP(@>(regulation(of(transcriptional(activity(by(pml(TP53,RB1);EP300,CREBBP(@>(E2F(transcription(factor(network(CDKN1B,CDKN2A,MCL1,RB1,E2F4);EP300,CRE0.574890625 0.02290476 CREBBP(@>(NCOA3;EP300(@>(TP53;TBX5(@>(HNF1A,TP53,TBX30.2405 CREBBP(@>(NCOA3;EP300(@>(TP53 Hereditary(Sensory(Neuropathy BRCA 1 0.035221138 NTRK1(@>(Trk(receptor(signaling(mediated(by(PI3K(and(PLC@gamma(CCND1,PIK3CA,PIK3R1,AKT1);HSPB1(@>(p38(mapk(signaling(pathway(MEF2A,MAP3K1,MAP2K4);HSPB1(@>(downregulated(of(mta@3(in(er@negative(breast(tumors(CDH1,MBD3);LMNA(@>(tnfr1(signaling(pathway(CASP8,MAP2K4);MED25(@>(Generic(Transcription(Pathway(MED12,MED15);NTRK1(@>(trka(receptor(signaling(pathway(PIK3CA,PIK3R1,AKT1);NDRG1(@>(Validated(targets(of(C@MYC(transcriptional(repression(ERBB2,CDKN1B,CCND1,ZFP36L1);NTRK1(@>(role(of(erk5(in(neuronal(survival(pathway(PIK3CA,MEF2A,PIK3R1,AKT1);EGR2(@>(IL4@mediated(signaling(events(AKT1,MYB,PIK3R1,PIK3CA);NTRK1(@>(p75(NTR)@mediated(signaling(AKT1,NDNL2,TP53,PIK3R1,PIK3CA);PMP22(@>(a6b1(and(a6b4(Integrin(signaling(CDH1,AKT1,ERBB2,PIK3R1,PIK3CA)0.403534273 1 1 Severe(Combined(Immunodeficiency BRCA 1 0.033379032 CD3D,PTPRC,ZAP70(@>(t(cell(receptor(signaling(pathway(NFATC1,MAP3K1,PIK3CA,PIK3R1,MAP2K4);JAK3,IL2RG(@>(IL2(signaling(events(mediated(by(PI3K(MYB,TERT,PIK3CA,PIK3R1,AKT1);PTPRC(@>(BCR(signaling(pathway(NFATC1,AKT1,PIK3CA,MAP3K1,PTEN,PIK3R1);CD3D(@>(the(co@stimulatory(signal(during(t@cell(activation(PIK3CA,PIK3R1);JAK3(@>(CD40/CD40L(signaling(AKT1,MAP3K1,PIK3CA,PIK3R1,MAP2K4);JAK3,IL2RG(@>(il@7(signal(transduction(PIK3CA,PIK3R1);JAK3,IL2RG(@>(IL4@mediated(signaling(events(AKT1,MYB,PIK3R1,PIK3CA);CD3D(@>(Immunoregulatory(interactions(between(a(Lymphoid(and(a(non@Lymphoid(cell(CDH1,HLA@B,HLA@A);ADA(@>(Validated(transcriptional(targets(of(deltaNp63(isoforms(CDKN2A,RUNX1,AXL)0.001276652 ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(SELPLG(1.40e@03);ZAP70,IL2RG,JAK3,PTPRC,DCLRE1C(@>(ZNF384(5.80e@03);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(MAP3K1(3.39e@03);IL2RG,ADA,RFXAP,DCLRE1C(@>(CBFB(2.74e@04);RFXAP,AK2,DCLRE1C(@>(MYB(1.73e@04);IL2RG,ADA,RFXAP,PNP,RFX5,AK2,DCLRE1C(@>(CCDC18(1.15e@04);RFXAP,DCLRE1C(@>(FNTA(4.47e@02);(@>(E2F4(9.86e@04);ZAP70,RFXANK,IL7R,CD3D(@>(RPL13(3.43e@03);NHEJ1,RFXANK(@>(RPL18(5.22e@03);(@>(GATA3(1.97e@02);ZAP70,IL2RG,JAK3,PNP,RFX5,PTPRC,IL7R,CD3D(@>(USP36(1.43e@02);ZAP70,IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(ZNF276(3.74e@03);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(KDM5A(9.26e@03);ZAP70,IL2RG,JAK3,RFXAP,CD3D,DCLRE1C(@>(CTCF(2.54e@02);IL2RG,JAK3,ADA,PNP,RFX5,PTPRC(@>(NFATC1(1.90e@02);ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(CDKN1B(1.09e@02);ZAP70,IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(HLA@A(7.19e@03);ZAP70,IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(DENND4B(3.72e@03);ZAP70,IL2RG,RFXAP,IL1 1 Specified(Hamartoses BRCA 0.7 PTEN 0.000162829 PTEN(@>(skeletal(muscle(hypertrophy(is(regulated(via(akt@mtor(pathway(PTEN,PIK3CA,IGF1R,PIK3R1,AKT1);PTEN(@>(regulation(of(eif@4e(and(p70s6(kinase(PTEN,AKT1,PIK3R1,PIK3CA);VHL(@>(vegf(hypoxia(and(angiogenesis(PIK3CA,PIK3R1,AKT1);PTEN(@>(mtor(signaling(pathway(AKT1,PIK3CA,PIK3R1,PTEN);PTEN(@>(BCR(signaling(pathway(NFATC1,AKT1,PIK3CA,MAP3K1,PTEN,PIK3R1);VHL(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53);PTEN(@>(RhoA(signaling(pathway(PTEN,CDKN1B,MAP2K4);PTEN(@>(pten(dependent(cell(cycle(arrest(and(apoptosis(PTEN,AKT1,PIK3CA,PIK3R1,CDKN1B);PTEN(@>(Negative(regulation(of(the(PI3K/AKT(network(PTEN,AKT1);PTEN(@>(Signaling(events(mediated(by(Stem(cell(factor(receptor((c@Kit)(PTEN,PIK3CA,PIK3R1,AKT1)0.736815818 0 PTEN(@>(IGF1R;SDHB(@>(CDKN2A;STK11(@>(TP531 Li(Fraumeni(and(Related(Syndromes BRCA 0.05 CDKN2A,TP53 8.57E@15 TP53(@>(chaperones(modulate(interferon(signaling(pathway(TP53,RB1);TP53(@>(apoptotic(signaling(in(response(to(dna(damage(AKT1,TP53);CDKN2A(@>(Validated(transcriptional(targets(of(AP1(family(members(Fra1(and(Fra2(CCND1,NFATC1,CDKN2A);TP53(@>(overview(of(telomerase(protein(component(gene(htert(transcriptional(regulation(TERT,TP53);CDKN2A(@>(Coregulation(of(Androgen(receptor(activity(CCND1,AKT1,CDKN2A,CASP8);TP53,CHEK2(@>(role(of(brca1(brca2(and(atr(in(cancer(susceptibility(TP53,FANCA);CDKN2A,TP53(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53);TP53(@>(telomeres(telomerase(cellular(aging(and(immortality(AKT1,TERT,TP53,RB1);CDKN2A(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(CDKN2A,TBL1XR1,CCND1,MED12,CDH1,TERT);TP53(@>(btg(family(proteins(and(cell(cycle(regulation(CCND1,TP53,RB1);TP53(@>(Transcriptional((activation(of((cell(cycle(inhibitor(p21(TP53);TP53(@>(p53(signaling(pathway(CCND1,TP53,RB1);TP53(@>(regulation(of(transcriptional(activity(by(pml(TP53,RB0.720244363 1 1 Lipoprotein(Deficiencies BRCA 1 0.890189985 0.077804012 (@>(KCNMB3(1.34e@02);(@>(UNC93A(1.98e@02);MTTP,APOB,LCAT,SAR1B,APOA1(@>(F10(1.16e@02);MTTP,APOB,LCAT,SAR1B,APOA1(@>(CYP2E1(1.16e@02);APOB,LCAT,APOA1(@>(SLC6A12(1.84e@02);MTTP,APOB,LCAT,SAR1B,APOA1,ABCA1(@>(NEU4(1.45e@02);MTTP,APOB,LCAT,SAR1B,APOA1(@>(HNF1A(1.45e@02);MTTP,APOB,LCAT,SAR1B,APOA1(@>(AQP11(2.84e@02)1 1 Disorders(of(Urea(Cycle(Metabolism BRCA 1 0.223337184 0.086959582 (@>(TMEM184A(4.36e@02);(@>(KCNMB3(1.76e@02);(@>(FOXA1(2.92e@02);ASS1,NAGS,ASL,CPS1(@>(HSBP1L1(1.85e@02);(@>(UNC93A(1.76e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(F10(3.36e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(CYP2E1(3.36e@02);ASS1(@>(TBX3(4.36e@02);NAGS,ARG1,ASL,CPS1(@>(SLC6A12(2.79e@02);ASS1,NAGS,ARG1,ASL(@>(ISOC2(2.57e@02);NAGS,ASL(@>(CDK10(4.25e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(HNF1A(1.49e@02);NAGS,ARG1(@>(AQP11(2.08e@02)1 1 Retinitis(Pigmentosa BRCA 1 RPGR 1 1.07E@13 TTC8,CERKL,FAM161A(@>(CHRFAM7A(2.17e@02);SNRNP200,CA4,MERTK,PDE6B,PRPF3,BEST1,RP2,TOPORS,FAM161A,SEMA4A(@>(VEZF1(1.76e@03);CA4,KLHL7,SPATA7,CRB1,FAM161A(@>(IGF1R(2.39e@02);CNGA1(@>(MUC20(4.95e@03);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,CRB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,CNGA1,RP1,PRCD,C2orf71,RPE65,SAG,RBP3,FAM161A,ABCA4(@>(SLC6A13(4.54e@13);CRX,LRAT,FSCN2,RDH12,SNRNP200,PRPH2,RHO,CNGB1,EYS,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,PROM1,C2orf71,RPE65,SAG,RBP3,FAM161A,ABCA4(@>(CROCC(8.93e@16);CRX,SNRNP200,TULP1(@>(CCDC144NL(8.18e@04)0.04182609 IMPDH1(@>(PABPC3;PRPF3(@>(RPL18;PRPF31(@>(RPL13,TXNL4A;PRPF8(@>(SF3B1,PABPC3;RHO(@>(SF3B1;SNRNP200(@>(TXNL4A,PABPC31 Haemophilia BRCA 1 0.052116948 F9(@>(Formation(of(Fibrin(Clot((Clotting(Cascade)(F10)1 0.702 1 Chronic(Granulomatous(Disease COAD 1 1 0.087493235 NCF2(@>(KRAS(4.54e@02);NCF2,CYBB,CYBA(@>(TCF7L2(1.51e@02);NCF2,NCF4,CYBB,CYBA(@>(PCBP1(1.53e@02);NCF2,NCF4,CYBB,CYBA(@>(FBRS(1.53e@02);NCF2,NCF4(@>(TNFRSF10C(1.54e@02);NCF2,NCF4,CYBB(@>(TRAPPC11(4.57e@02);NCF2,CYBB(@>(ACVR2A(4.65e@02);NCF2,NCF4,CYBB,CYBA(@>(GGT1(4.40e@02);NCF2,NCF4,CYBB,CYBA(@>(BRAF(2.90e@02)1 1 Polycystic(Kidney,(Autosomal(Dominant COAD 1 1.58E@05 TSC2(@>(LKB1(signaling(events(SMAD4,TP53);TSC2(@>(mTOR(signaling(pathway(NRAS,BRAF,KRAS);TSC2(@>(Validated(targets(of(C@MYC(transcriptional(repression(SMAD4,SMAD2,SMAD3)0.807669883 1 0.26722222 Inherited(Anomalies(of(the(Skin COAD 1 2.64E@06 TERT(@>(telomeres(telomerase(cellular(aging(and(immortality(TP53,KRAS);KRT1,TERT(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(APC,CDKN2A,TCF7L2);TERT(@>(Validated(targets(of(C@MYC(transcriptional(activation(SMAD4,TP53,SMAD3)0.147886703 1 1 Spinocerebellar(Ataxia COAD 1 0.380460894 0.003170856 ATXN7,TDP1,SYNE1,ATM,TTBK2,TBP,ITPR1,PPP2R2B(@>(FHIT(3.79e@03);JPH3,CACNA1A,ATXN2,SYNE1,SPTBN2,TTBK2,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(PLK5(2.24e@04);ZNF592,ATXN7,TDP1,SYNE1,ATM,TTBK2,TBP,ITPR1,SETX,ATXN1(@>(PIK3CA(8.50e@03);JPH3,ZNF592,CACNA1A,ATXN2,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B,ATXN1(@>(CCSER1(1.43e@03);JPH3,ZNF592,CACNA1A,ATXN7,ATXN2,PDYN,TDP1,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SETX,SYT14,PPP2R2B,ATXN1(@>(WHSC1L1(2.24e@04);JPH3,ZNF592,CACNA1A,ATXN2,PDYN,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,AFG3L2,SYT14,PPP2R2B(@>(PARK2(1.71e@03);JPH3,ZNF592,CACNA1A,ATXN2,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,PRNP,ITPR1,PRKCG,FGF14,SETX,SYT14,PPP2R2B,ATXN1(@>(ZC3H13(6.35e@04);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(DLGAP1(1.67e@03);ATM(@>(FNTA(3.08e@02);JPH3,ATXN2,SPTBN2,TTBK2,KCNC3,PPP2R2B(@>(ENPP6(7.78e@03);JPH3,APTX,CACNA1A,ATXN7,ATXN2,PDYN,TDP1,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,PRNP,ITPR1,PRKCG,1 1 Combined(Heart(and(Skeletal(Defects COAD 1 0.00325544 EP300(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(APC,CDKN2A,TCF7L2);EP300(@>(Validated(targets(of(C@MYC(transcriptional(repression(SMAD4,SMAD2,SMAD3);EP300,CREBBP(@>(Validated(targets(of(C@MYC(transcriptional(activation(SMAD4,TP53,SMAD3);CREBBP(@>(Signaling(events(mediated(by(TCPTP(PIK3CA,INS)0.664784536 0.05772 CREBBP(@>(TP53,UBE3A;EP300(@>(TP531 Neurofibromatosis COAD 1 0.012213067 NF1(@>(Regulation(of(Ras(family(activation(NRAS,KRAS)1 0.55985246 0.53565909 Hereditary(Sensory(Neuropathy COAD 1 0.038741758 NTRK1(@>(Trk(receptor(signaling(mediated(by(PI3K(and(PLC@gamma(PIK3CA,NRAS,KRAS);NTRK1(@>(ARMS@mediated(activation(BRAF);NDRG1(@>(Validated(targets(of(C@MYC(transcriptional(repression(SMAD4,SMAD2,SMAD3);NTRK1(@>(Frs2@mediated(activation(BRAF);NTRK1(@>(Signalling(to(p38(via(RIT(and(RIN(BRAF)0.718504577 1 1 Severe(Combined(Immunodeficiency COAD 1 0.461439204 0.001993726 ZAP70,IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(B2M(8.02e@03);ZAP70,IL2RG,JAK3,RFXAP,IL7R,CD3D,DCLRE1C(@>(CSTF2T(1.17e@02);IL2RG,JAK3,RFXANK,RFX5,PTPRC(@>(PRR14(6.59e@03);IL2RG,JAK3,PNP,PTPRC(@>(PCBP1(4.79e@02);ZAP70,IL2RG,JAK3,PTPRC,CD3D,DCLRE1C(@>(SMAD2(7.76e@03);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(IKZF3(8.72e@05);ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(ZNF18(8.78e@03);(@>(NRAS(4.87e@02)1 1 Specified(Hamartoses COAD 1 0.129885453 0.682837057 0 PTEN(@>(PIK3CA;SDHB(@>(CDKN2A;STK11(@>(TP531 Li(Fraumeni(and(Related(Syndromes COAD 0.03 CDKN2A,TP53 1.64E@21 CDKN2A(@>(C@MYC(pathway(CDKN2A,FBXW7);TP53(@>(LKB1(signaling(events(SMAD4,TP53);CDKN2A,TP53(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53);TP53(@>(telomeres(telomerase(cellular(aging(and(immortality(TP53,KRAS);CDKN2A(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(APC,CDKN2A,TCF7L2);TP53(@>(Transcriptional((activation(of((cell(cycle(inhibitor(p21(TP53);TP53,CHEK2(@>(PLK3(signaling(events(TP53);TP53(@>(Validated(targets(of(C@MYC(transcriptional(activation(SMAD4,TP53,SMAD3);CDKN2A(@>(Validated(transcriptional(targets(of(deltaNp63(isoforms(CDKN2A,FBXW7)0.700494843 0.1924 CHEK2(@>(TP53,SMAD4;TP53(@>(UBE3A1 Retinitis(Pigmentosa COAD 1 0.968925499 0.001527057 CA4,SPATA7,CRB1,CERKL(@>(PLK5(1.76e@02);(@>(IMMP2L(5.09e@05);SPATA7,CRB1,MERTK,PDE6B,PRCD,FAM161A(@>(ENPP6(2.88e@03)1 0.64133333 Hereditary(Hemorrhagic(Telangiectasia COAD 0.46 SMAD4 3.56E@24 SMAD4(@>(TGF@beta(receptor(signaling(SMAD4,SMAD2,SMAD3);SMAD4,ACVRL1,ENG(@>(ALK1(signaling(events(SMAD4,ACVR2A,ID1);SMAD4(@>(LKB1(signaling(events(SMAD4,TP53);SMAD4(@>(Signaling(by(BMP(SMAD4);SMAD4(@>(Signaling(by(TGF(beta(SMAD4);SMAD4(@>(Regulation(of(cytoplasmic(and(nuclear(SMAD2/3(signaling(SMAD4,SMAD2,SMAD3);SMAD4(@>(Validated(targets(of(C@MYC(transcriptional(repression(SMAD4,SMAD2,SMAD3);SMAD4(@>(Validated(targets(of(C@MYC(transcriptional(activation(SMAD4,TP53,SMAD3)0.518227346 0 SMAD4(@>(ACVR2A 1 Dopa@Responsive(Dystonia COAD 0.53 TH 0.000139524 TH(@>(Alpha@synuclein(signaling(TH,PARK2) 1 1 1 Chronic(Granulomatous(Disease GBM 1 1 0.07356062 NCF2,NCF4,CYBB,CYBA(@>(PTEN(1.48e@02);NCF2,NCF4,CYBB,CYBA(@>(ZZEF1(1.72e@02);NCF4,CYBA(@>(CDKN1B(4.22e@02);(@>(C12orf5(1.72e@02);NCF2,NCF4,CYBB,CYBA(@>(HRH2(9.80e@03);NCF2(@>(TNFRSF9(1.98e@02);NCF2,NCF4,CYBB(@>(STAG2(2.52e@02);NCF2,NCF4,CYBB,CYBA(@>(PRDM2(4.38e@02);NCF2,NCF4,CYBB,CYBA(@>(RAP1B(1.23e@02);NCF2,CYBB,CYBA(@>(METTL9(1.10e@02);NCF2,NCF4,CYBB,CYBA(@>(LPAR6(1.32e@02);NCF2,NCF4,CYBB,CYBA(@>(CAMK1D(1.56e@02);NCF2,NCF4,CYBB,CYBA(@>(IGSF6(3.92e@02);NCF2,NCF4,CYBB,CYBA(@>(RB1(1.78e@02);NCF2,NCF4,CYBB,CYBA(@>(SCIMP(2.19e@02);NCF2,NCF4,CYBB,CYBA(@>(EVI2A(1.49e@02);NCF2,NCF4,CYBB,CYBA(@>(EVI2B(1.30e@02);NCF2,NCF4,CYBB,CYBA(@>(BRAF(2.52e@02)1 1 Cerebral(Degeneration(Due(to(Generalized(LipidosesGBM 1 0.039068638 SMPD1(@>(IL2(signaling(events(mediated(by(PI3K(AKT1,PIK3R1,PIK3CA);SMPD1(@>(phospholipids(as(signalling(intermediaries(AKT1,PDGFRA,PIK3R1,PIK3CA)0.165898202 1 1 Congenital(Ichthyosis GBM 1 0.741625429 0.024789969 (@>(CDKN2B(2.27e@03);ALOX12B,CSTA(@>(ADSSL1(4.30e@02);SPINK5,CSTA(@>(SVIL(3.26e@02);ABCA12,TGM1(@>(EGFR(4.30e@02)0.8658 1 Inherited(Adrenogenital(Disorders GBM 1 0.521588572 0.097166245 HSD3B2,POR,CYP17A1,CYP21A2(@>(ITIH1(1.83e@02)1 1 Pervasive,(Specified(Congenital(Anomalies GBM 1 BRAF 2.63E@31 NRAS,HRAS,RAF1,SOS1,KRAS(@>(EGFR(Inhibitor(Pathway,(Pharmacodynamics(PIK3CA,EGFR,CAMK1D,PIK3R1,AKT1);HRAS,SOS1(@>(nfat(and(hypertrophy(of(the(heart((AKT1,PIK3R1,PIK3CA);HRAS,NRAS,SOS1,KRAS(@>(Trk(receptor(signaling(mediated(by(PI3K(and(PLC@gamma(PIK3CA,PIK3R1,AKT1);HRAS,RAF1,SOS1(@>(pdgf(signaling(pathway(PIK3CA,PDGFRA,PIK3R1);PTCH1(@>(Signaling(events(mediated(by(the(Hedgehog(family(AKT1,PIK3R1,PIK3CA);SOS1(@>(PDGFR@alpha(signaling(pathway(PIK3CA,PDGFRA,PIK3R1);HRAS,SOS1(@>(vegf(hypoxia(and(angiogenesis(PIK3CA,PIK3R1,AKT1);PTPN11(@>(IFN@gamma(pathway(AKT1,RAP1B,PIK3R1,PIK3CA);PTPN11,SOS1(@>(IL2(signaling(events(mediated(by(STAT5(PIK3CA,CCND2,CDK6,PIK3R1);PTPN11(@>(VEGFR1(specific(signals(AKT1,PIK3R1,PIK3CA);HRAS(@>(LPA(receptor(mediated(events(AKT1,EGFR,PIK3R1);HRAS,RAF1(@>(Nongenotropic(Androgen(signaling(AKT1,PIK3R1,PIK3CA);PTPN11,SOS1(@>(IL2(signaling(events(mediated(by(PI3K(AKT1,PIK3R1,PIK3CA);HRAS,RAF1,PTPN11,SOS1(@>(Fc@epsilon(receptor(I(signaling(in(mast(cells(AKT1,PIK3R1,PIK3CA);HRAS,RAF1,SOS1(@>(tpo0.115677047 0 BRAF(@>(EGFR,DGKZ;HRAS(@>(BRAF,EGFR;KRAS(@>(RAP1B;NRAS(@>(BRAF,PIK3CA;PTPN11(@>(PTEN,EGFR,RB1,STAG2,STAG20.2405 BRAF(@>(AKT1,ARID2,TP53,DGKZ;CHD7(@>(TP53;CUL7(@>(NF1;HRAS(@>(PIK3CA;KRAS(@>(BRAF;NIPBL(@>(PIK3R1,PIK3CA,CDK6,MDM2,CD33;NRAS(@>(EGFR;PTPN11(@>(PIK3R1,RB1;RAF1(@>(AKT1;RPS6KA3(@>(BRAF,EGFR,STAG2;SMC1A(@>(CDK4,STAG2;SMC3(@>(CDK4,MET;SOS1(@>(EGFR;TRIM32(@>(PIK3R1,MYCN Diamond@Blackfan(Anemia GBM 0.7 RPL5 1 0.011511663 RPS26,RPS19,RPS10,RPS7(@>(SIVA1(1.42e@03);RPS26,RPS24,RPS7(@>(TFB2M(2.01e@03);RPS26(@>(CDKN2A(1.64e@02);RPS24,RPS10,RPL11,RPL35A(@>(RPL5(7.91e@04);(@>(C12orf5(4.18e@02);(@>(CCNE1(1.34e@02);RPS26,RPS19,RPS7(@>(TP53(2.74e@03);RPS26(@>(CDK4(1.65e@02);RPS26,RPS19,RPL35A,RPS7(@>(CNIH4(2.94e@03)1 0.26054167 Inherited(Anomalies(of(the(Skin GBM 1 1.50E@05 ATP2A2(@>(nfat(and(hypertrophy(of(the(heart((AKT1,PIK3R1,PIK3CA);TERT(@>(IL2(signaling(events(mediated(by(PI3K(AKT1,PIK3R1,PIK3CA);TERT(@>(telomeres(telomerase(cellular(aging(and(immortality(AKT1,TP53,RB1);TERT(@>(role(of(nicotinic(acetylcholine(receptors(in(the(regulation(of(apoptosis(PIK3CA,PIK3R1,AKT1);TINF2,TERT,DKC1(@>(Regulation(of(Telomerase(CDKN1B,EGFR,AKT1)0.165069058 1 1 Spinocerebellar(Ataxia GBM 1 0.007540264 PRKCG(@>(EGFR(Inhibitor(Pathway,(Pharmacodynamics(PIK3CA,EGFR,CAMK1D,PIK3R1,AKT1);ATM(@>(apoptotic(signaling(in(response(to(dna(damage(AKT1,TP53);ATM(@>(p53(pathway(AKT1,CDKN2A,RPL5,TP53,MDM2);ATM(@>(cell(cycle:(g2/m(checkpoint(MDM2,TP53);CACNA1A(@>(Anti@diabetic(Drug(Potassium(Channel(Inhibitors(Pathway,(Pharmacodynamics(AKT1,PIK3R1,PIK3CA);ATM(@>(BARD1(signaling(events(CCNE1,TP53);ATM(@>(atm(signaling(pathway(MDM2,TP53);ATM(@>(rb(tumor(suppressor/checkpoint(signaling(in(response(to(dna(damage(CDK4,TP53,RB1);ATM(@>(E2F(transcription(factor(network(CDKN1B,CDKN2A,CCNE1,RB1,CDKN2C);CACNA1A(@>(rac1(cell(motility(signaling(pathway(PIK3CA,PDGFRA,PIK3R1);ATM(@>(hypoxia(and(p53(in(the(cardiovascular(system(AKT1,MDM2,TP53);ATM(@>(Regulation(of(Telomerase(CDKN1B,EGFR,AKT1)0.003470279 POLG,ATXN7,TDP1,ATM,TBP,NOP56,AFG3L2,C10orf2(@>(POLR3E(2.52e@02);POLG,ATXN7,TDP1,ATM,TTBK2,TBP,ITPR1(@>(MARCH9(5.79e@03);ZNF592,ATXN7,TDP1,SYNE1,ATM,TTBK2,TBP,ITPR1,SETX,ATXN1(@>(PIK3CA(9.40e@03);JPH3,APTX,ZNF592,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,TBP,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(NF1(1.85e@03);JPH3,ZNF592,ATXN2,SPTBN2,TTBK2,KCNC3,ITPR1,PRKCG,FGF14,PPP2R2B(@>(PIK3C2B(6.44e@03);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(FGFR3(3.99e@03);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(BRSK2(1.51e@03);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(LSAMP(1.98e@03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(NETO1(1.98e@03);JPH3,CACNA1A,TDP1,SYNE1,ATM,TTBK2,TBP,KCNC3,ITPR1,FGF14(@>(ELP4(1.93e@03);TTBK2,AFG3L2(@>(UQCRC2(4.16e@02);JPH3,CACNA1A,ATXN7,TDP1,SYNE1,ATM,SPTBN2,TTBK2,T1 1 Hypopituitarism GBM 1 8.68E@11 GH1(@>(regulation(of(eif@4e(and(p70s6(kinase(PTEN,AKT1,PIK3R1,PIK3CA);GLI2(@>(Signaling(events(mediated(by(the(Hedgehog(family(AKT1,PIK3R1,PIK3CA);GH1(@>(mtor(signaling(pathway(AKT1,PIK3CA,PIK3R1,PTEN);BTK(@>(Fc@epsilon(receptor(I(signaling(in(mast(cells(AKT1,PIK3R1,PIK3CA);BTK(@>(BCR(signaling(pathway(AKT1,VAV2,PIK3CA,PIK3R1,PTEN);FGFR1(@>(Syndecan@3@mediated(signaling(events(EGFR,FGFR3);FGFR1(@>(Signal(transduction(by(L1(VAV2,EGFR);GH1(@>(growth(hormone(signaling(pathway(PIK3CA,PIK3R1);GH1(@>(akt(signaling(pathway(PIK3CA,PIK3R1,AKT1);BTK(@>(phosphoinositides(and(their(downstream(targets(AKT1,VAV2);GH1(@>(trefoil(factors(initiate(mucosal(healing(AKT1,EGFR,PIK3R1,PIK3CA);FGFR1(@>(FGF(signaling(pathway(AKT1,FGFR3,PIK3CA,PIK3R1,MET)1 0.26236364 1 Combined(Heart(and(Skeletal(Defects GBM 1 6.38E@05 CREBBP(@>(nfat(and(hypertrophy(of(the(heart((AKT1,PIK3R1,PIK3CA);EP300,CREBBP(@>(IFN@gamma(pathway(AKT1,RAP1B,PIK3R1,PIK3CA);EP300,CREBBP(@>(Direct(p53(effectors(TP53,MET,EGFR,C12orf5,RB1,PTEN,MDM2);EP300,CREBBP(@>(p53(pathway(AKT1,CDKN2A,RPL5,TP53,MDM2);EP300(@>(cell(cycle:(g2/m(checkpoint(MDM2,TP53);CREBBP(@>(regulation(of(transcriptional(activity(by(pml(TP53,RB1);EP300,CREBBP(@>(E2F(transcription(factor(network(CDKN1B,CDKN2A,CCNE1,RB1,CDKN2C);EP300,CREBBP(@>(il@7(signal(transduction(PIK3CA,PIK3R1);EP300(@>(hypoxia(and(p53(in(the(cardiovascular(system(AKT1,MDM2,TP53);CREBBP(@>(Signaling(events(mediated(by(Stem(cell(factor(receptor((c@Kit)(PTEN,PIK3CA,AKT1,PIK3R1,PIK3C2B);EP300(@>(ATF@2(transcription(factor(network(PDGFRA,CDK4,NF1,RB1);CREBBP(@>(Signaling(events(mediated(by(TCPTP(MET,EGFR,PIK3R1,PIK3CA);EP300,CREBBP(@>(FOXM1(transcription(factor(network(CDKN2A,CDK4,CCNE1,RB1)0.662181163 0.10307143 CREBBP(@>(CTBP1;EP300(@>(TP53,TP531 Specified(Anomalies(of(the(Musculoskeletal(SystemGBM 0.72 FGFR3 0.013251879 SNAI2(@>(Direct(p53(effectors(TP53,MET,EGFR,C12orf5,RB1,PTEN,MDM2);FGFR3(@>(Syndecan@3@mediated(signaling(events(EGFR,FGFR3);FGFR3(@>(FGFR3b(ligand(binding(and(activation(FGFR3);FGFR3(@>(FGFR3c(ligand(binding(and(activation(FGFR3);MITF,SNAI2(@>(Signaling(events(mediated(by(Stem(cell(factor(receptor((c@Kit)(PTEN,PIK3CA,AKT1,PIK3R1,PIK3C2B);MITF(@>(IL6@mediated(signaling(events(AKT1,PIK3R1,PIK3CA);FGFR3(@>(FGF(signaling(pathway(AKT1,FGFR3,PIK3CA,PIK3R1,MET)0.867886984 1 1 Neurofibromatosis GBM 0.44 NF1 0.003812225 NF1(@>(ATF@2(transcription(factor(network(PDGFRA,CDK4,NF1,RB1)0.689368988 0 NF1(@>(EVI2A,CDKN2A;NF2(@>(NF11 Hereditary(Sensory(Neuropathy GBM 1 INF2 0.097855772 NTRK1(@>(Trk(receptor(signaling(mediated(by(PI3K(and(PLC@gamma(PIK3CA,PIK3R1,AKT1);NDRG1(@>(Direct(p53(effectors(TP53,MET,EGFR,C12orf5,RB1,PTEN,MDM2);NTRK1(@>(trka(receptor(signaling(pathway(PIK3CA,PIK3R1,AKT1);NTRK1(@>(role(of(erk5(in(neuronal(survival(pathway(PIK3CA,PIK3R1,AKT1);NTRK1(@>(Signalling(to(p38(via(RIT(and(RIN(BRAF);EGR2(@>(IL4@mediated(signaling(events(AKT1,PIK3R1,PIK3CA);NTRK1(@>(p75(NTR)@mediated(signaling(AKT1,OMG,TP53,PIK3R1,PIK3CA);PMP22(@>(a6b1(and(a6b4(Integrin(signaling(AKT1,MET,EGFR,PIK3R1,PIK3CA)0.381135563 1 1 Tuberous(Sclerosis GBM 1 0.004334094 TSC2(@>(AKT(phosphorylates(targets(in(the(cytosol(AKT1,MDM2);TSC2,TSC1(@>(mtor(signaling(pathway(AKT1,PIK3CA,PIK3R1,PTEN);TSC2(@>(Direct(p53(effectors(TP53,MET,EGFR,C12orf5,RB1,PTEN,MDM2);TSC2,TSC1(@>(mTOR(signaling(pathway(EEF2K,AKT1,CCNE1,BRAF)0.686334259 0.63004225 0.180375 TSC1(@>(CDKN1B;TSC2(@>(AKT1,CDKN1B Severe(Combined(Immunodeficiency GBM 1 0.000972189 JAK3,IL2RG(@>(IL2(signaling(events(mediated(by(STAT5(PIK3CA,CCND2,CDK6,PIK3R1);JAK3,IL2RG(@>(IL2(signaling(events(mediated(by(PI3K(AKT1,PIK3R1,PIK3CA);PTPRC(@>(BCR(signaling(pathway(AKT1,VAV2,PIK3CA,PIK3R1,PTEN);CD3D(@>(the(co@stimulatory(signal(during(t@cell(activation(PIK3CA,PIK3R1);JAK3(@>(CD40/CD40L(signaling(AKT1,PIK3R1,PIK3CA);JAK3,IL2RG(@>(il@7(signal(transduction(PIK3CA,PIK3R1);JAK3,IL2RG(@>(IL4@mediated(signaling(events(AKT1,PIK3R1,PIK3CA);JAK3(@>(Signaling(events(mediated(by(TCPTP(MET,EGFR,PIK3R1,PIK3CA);CD3D,PTPRC(@>(CXCR4@mediated(signaling(events(PTEN,AKT1,RAP1B,PIK3R1,PIK3CA)0.001527057 IL2RG,PNP,PTPRC,DCLRE1C(@>(CD33(4.61e@03);ZAP70,ADA,RFXANK,CD3D(@>(SIVA1(6.96e@03);ZAP70,IL2RG,JAK3,RFXAP,IL7R,CD3D(@>(POLR3E(1.28e@02);ZAP70,IL2RG,JAK3,RFXAP,RFX5,IL7R,CD3D,DCLRE1C(@>(MARCH9(1.12e@02);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(ZZEF1(3.17e@02);ZAP70,RFXAP,IL7R,CD3D,DCLRE1C(@>(ELP4(3.17e@02);ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(CDKN1B(1.14e@02);ZAP70,IL2RG,NHEJ1,RFX5,IL7R,CD3D,DCLRE1C(@>(RPL5(5.40e@05);IL2RG,JAK3,PTPRC,DCLRE1C(@>(HRH2(3.56e@02);IL2RG,CIITA,PNP,RFX5(@>(TNFRSF9(1.40e@02);IL2RG,JAK3,ADA,RFXAP,PNP,PTPRC,DCLRE1C(@>(STAG2(1.27e@02);ZAP70,IL2RG,JAK3,RFXAP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C,RAG2,RAG1(@>(ARID2(1.14e@02);IL2RG,CIITA,ADA,PNP,PTPRC(@>(RAP1B(4.05e@02);ADA,AK2,DCLRE1C(@>(CDK6(3.58e@04);ZAP70,IL2RG,JAK3,PTPRC,IL7R,CD3D,DCLRE1C(@>(LPAR6(6.96e@03);IL2RG,CIITA,JAK3,ADA,PNP,RFX5,PTPRC,DCLRE1C(@>(RB1(3.56e@03);IL2RG,CIITA,RFX5,PTPRC,DCLRE1C(@>(SCIMP(4.22e@03);ZAP70,IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(EVI2A(1.27e@02);IL2RG,JAK3,RFX5,PTPRC,DCLRE1C1 1 Specified(Hamartoses GBM 0.62 PTEN 3.91E@09 PTEN(@>(skeletal(muscle(hypertrophy(is(regulated(via(akt@mtor(pathway(PTEN,PIK3CA,PIK3R1,AKT1);PTEN(@>(regulation(of(eif@4e(and(p70s6(kinase(PTEN,AKT1,PIK3R1,PIK3CA);VHL(@>(vegf(hypoxia(and(angiogenesis(PIK3CA,PIK3R1,AKT1);PTEN(@>(mtor(signaling(pathway(AKT1,PIK3CA,PIK3R1,PTEN);PTEN(@>(Direct(p53(effectors(TP53,MET,EGFR,C12orf5,RB1,PTEN,MDM2);PTEN(@>(BCR(signaling(pathway(AKT1,VAV2,PIK3CA,PIK3R1,PTEN);VHL(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53);PTEN(@>(pten(dependent(cell(cycle(arrest(and(apoptosis(PTEN,AKT1,PIK3CA,PIK3R1,CDKN1B);PTEN(@>(Negative(regulation(of(the(PI3K/AKT(network(PTEN,AKT1);PTEN(@>(Signaling(events(mediated(by(Stem(cell(factor(receptor((c@Kit)(PTEN,PIK3CA,AKT1,PIK3R1,PIK3C2B);PTEN(@>(CXCR4@mediated(signaling(events(PTEN,AKT1,RAP1B,PIK3R1,PIK3CA)0.451482314 0 PTEN(@>(AKT1;SDHD(@>(EGFR;STK11(@>(CDKN2A1 Li(Fraumeni(and(Related(Syndromes GBM 0.03 CDKN2A,TP53 4.83E@30 TP53(@>(Aurora(A(signaling(AKT1,MDM2,TP53);TP53(@>(chaperones(modulate(interferon(signaling(pathway(TP53,RB1);TP53(@>(apoptotic(signaling(in(response(to(dna(damage(AKT1,TP53);TP53(@>(Direct(p53(effectors(TP53,MET,EGFR,C12orf5,RB1,PTEN,MDM2);TP53,CDKN2A,CHEK2(@>(p53(pathway(AKT1,CDKN2A,RPL5,TP53,MDM2);TP53,CHEK2(@>(cell(cycle:(g2/m(checkpoint(MDM2,TP53);TP53(@>(estrogen(responsive(protein(efp(controls(cell(cycle(and(breast(tumors(growth(CDK4,TP53,CDK6);CDKN2A(@>(Coregulation(of(Androgen(receptor(activity(AKT1,CDKN2A,SVIL,CDK6);CDKN2A,TP53(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53);TP53(@>(telomeres(telomerase(cellular(aging(and(immortality(AKT1,TP53,RB1);TP53(@>(BARD1(signaling(events(CCNE1,TP53);TP53(@>(btg(family(proteins(and(cell(cycle(regulation(TP53,RB1);TP53(@>(Transcriptional((activation(of((cell(cycle(inhibitor(p21(TP53);TP53,CHEK2(@>(atm(signaling(pathway(MDM2,TP53);TP53,CHEK2(@>(PLK3(signaling(events(CCNE1,TP53);TP53(@>(p53(signaling(pathway(CDK4,MDM2,CCNE1,TP53,RB1);TP50.665220849 0 CDKN2A(@>(CDKN2C;CHEK2(@>(CCND2;TP53(@>(PTEN0.2405 CDKN2A(@>(MYCN;CHEK2(@>(TP53;TP53(@>(CCND2 Lipoprotein(Deficiencies GBM 1 0.930390184 0.052743895 APOB,SAR1B(@>(IDH1(1.66e@02);MTTP,APOB,LCAT,SAR1B,APOA1(@>(ITIH1(5.60e@03);MTTP,APOB,LCAT,SAR1B,APOA1(@>(DMRTA1(1.54e@02)1 1 Disorders(of(Urea(Cycle(Metabolism GBM 1 0.086191281 ARG1(@>(IL4@mediated(signaling(events(AKT1,PIK3R1,PIK3CA);ARG1(@>(ATF@2(transcription(factor(network(PDGFRA,CDK4,NF1,RB1)0.093008554 NAGS,ASL(@>(VAV2(3.02e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(ITIH1(1.71e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(DMRTA1(1.71e@02)1 1 Retinitis(Pigmentosa GBM 1 0.439405616 3.62E@10 SNRNP200,ZNF513,CRB1,PDE6B,RP9,FAM161A(@>(MARCH9(8.30e@03);KLHL7,SPATA7,CRB1,PDE6B,FAM161A(@>(NETO1(9.25e@03);TTC8,SNRNP200(@>(MYCN(5.63e@03);(@>(TRH(1.81e@04);IDH3B(@>(ADSSL1(4.55e@02);RPGR,RP2,SEMA4A(@>(METTL9(4.57e@02);(@>(CDR2(1.43e@02);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,RP1,PRCD,RPE65,SAG,RBP3,ABCA4(@>(KCNJ13(6.03e@12)1 1 Chondrodystrophy GBM 0.47 FGFR3 0.0001457 FGFR3(@>(Syndecan@3@mediated(signaling(events(EGFR,FGFR3);FGFR3(@>(FGFR3b(ligand(binding(and(activation(FGFR3);FGFR3(@>(FGFR3c(ligand(binding(and(activation(FGFR3);FGFR3(@>(FGF(signaling(pathway(AKT1,FGFR3,PIK3CA,PIK3R1,MET)1 1 1 Osteogenesis(Imperfecta GBM 1 1 0.078898011 CRTAP,LEPRE1(@>(AKT1(1.95e@02);CRTAP,LEPRE1(@>(CDKN2B(1.15e@02);(@>(MDM2(3.45e@02);COL1A2,COL1A1(@>(KLHL9(1.59e@02);(@>(TRH(3.95e@02);(@>(SMYD3(3.49e@02);(@>(TSPAN31(1.57e@02);CRTAP,LEPRE1(@>(INF2(3.52e@02);LEPRE1(@>(MDK(1.92e@02);COL1A2(@>(WNT2(1.12e@02);(@>(CDK4(2.07e@02);CRTAP,COL1A2,COL1A1,LEPRE1(@>(CDH13(2.03e@02);CRTAP,COL1A2,COL1A1,LEPRE1(@>(PTPN21(1.38e@02);CRTAP,LEPRE1(@>(CNIH4(1.64e@02);COL1A2,COL1A1,LEPRE1(@>(CDR2(1.55e@02)1 1 Anophthalmos/Micropthalmos GBM 1 1 0.066215584 BMP4,STRA6,SOX2,RAX(@>(MDK(7.73e@03)1 1 Long(QT(Syndrome KICH 1 1 0.045450368 SCN5A,CACNA1C(@>(PXDNL(4.73e@03);CACNA1C,CAV3(@>(FBXL22(2.95e@02)1 1 Chronic(Granulomatous(Disease KICH 1 1 0.088531104 NCF2,NCF4,CYBB,CYBA(@>(USP3(1.59e@02);NCF2,NCF4,CYBB,CYBA(@>(PTEN(1.76e@02);NCF2,NCF4,CYBB,CYBA(@>(RB1CC1(1.86e@02)1 1 Polycystic(Kidney,(Autosomal(Dominant KICH 1 1.61E@09 TSC2(@>(mtor(signaling(pathway(PTEN);TSC2(@>(Direct(p53(effectors(PTEN,TP53);TSC2(@>(LKB1(signaling(events(TP53);TSC2(@>(mTOR(signaling(pathway(RB1CC1)1 1 1 Spinocerebellar(Ataxia KICH 1 1.66E@10 ATM(@>(apoptotic(signaling(in(response(to(dna(damage(TP53);ATM(@>(regulation(of(cell(cycle(progression(by(plk3(TP53);ATM(@>(p53(pathway(TP53);ATM(@>(cell(cycle:(g2/m(checkpoint(TP53);ATM(@>(Autodegradation(of(the(E3(ubiquitin(ligase(COP1(TP53);ATM(@>(role(of(brca1(brca2(and(atr(in(cancer(susceptibility(TP53);ATM(@>(BARD1(signaling(events(TP53);ATM(@>(atm(signaling(pathway(TP53);ATM(@>(rb(tumor(suppressor/checkpoint(signaling(in(response(to(dna(damage(TP53);ATM(@>(hypoxia(and(p53(in(the(cardiovascular(system(TP53);TBP(@>(Glucocorticoid(receptor(regulatory(network(TP53)0.021413248 JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(SNTG1(1.92e@03);JPH3,ZNF592,CACNA1A,POLG,ATXN7,TDP1,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SETX,PPP2R2B,ATXN1(@>(HERC1(1.92e@03);JPH3,APTX,CACNA1A,ATXN2,SYNE1,SPTBN2,TTBK2,KCNC3,PRKCG,FGF14,SYT14,PPP2R2B,ATXN1(@>(ST18(3.88e@03);JPH3,ZNF592,ATXN7,SYNE1,ATM,SPTBN2,TTBK2,KCNC3,ITPR1,SETX,PPP2R2B,ATXN1(@>(RB1CC1(1.37e@02)1 1 Combined(Heart(and(Skeletal(Defects KICH 1 0.000738708 EP300,CREBBP(@>(Direct(p53(effectors(PTEN,TP53);EP300,CREBBP(@>(p53(pathway(TP53);EP300(@>(cell(cycle:(g2/m(checkpoint(TP53);CREBBP(@>(regulation(of(transcriptional(activity(by(pml(TP53);EP300(@>(hypoxia(and(p53(in(the(cardiovascular(system(TP53);EP300,CREBBP(@>(Glucocorticoid(receptor(regulatory(network(TP53);CREBBP(@>(Signaling(events(mediated(by(Stem(cell(factor(receptor((c@Kit)(PTEN);EP300,CREBBP(@>(Signaling(events(mediated(by(HDAC(Class(III(TP53)0.448457332 0 CREBBP(@>(TP53,TP53 1 Tuberous(Sclerosis KICH 1 1.61E@09 TSC2,TSC1(@>(mtor(signaling(pathway(PTEN);TSC2(@>(Direct(p53(effectors(PTEN,TP53);TSC2,TSC1(@>(LKB1(signaling(events(TP53);TSC2,TSC1(@>(mTOR(signaling(pathway(RB1CC1)0.839245915 1 1 Severe(Combined(Immunodeficiency KICH 1 1 0.07356062 ZAP70,IL2RG,CIITA,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(USP3(9.65e@03)1 1 Specified(Hamartoses KICH 0.13 PTEN 8.15E@35 PTEN(@>(skeletal(muscle(hypertrophy(is(regulated(via(akt@mtor(pathway(PTEN);PTEN(@>(regulation(of(eif@4e(and(p70s6(kinase(PTEN);PTEN(@>(Downstream(TCR(signaling(PTEN);PTEN(@>(mtor(signaling(pathway(PTEN);PTEN(@>(Direct(p53(effectors(PTEN,TP53);STK11(@>(LKB1(signaling(events(TP53);PTEN(@>(BCR(signaling(pathway(PTEN);VHL(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(TP53);PTEN(@>(RhoA(signaling(pathway(PTEN);PTEN(@>(pten(dependent(cell(cycle(arrest(and(apoptosis(PTEN);PTEN(@>(Negative(regulation(of(the(PI3K/AKT(network(PTEN);PTEN(@>(Signaling(events(mediated(by(Stem(cell(factor(receptor((c@Kit)(PTEN);PTEN(@>(TCR(signaling(in(naï(ve(CD4+(T(cells(PTEN)1 0.0925 PTEN(@>(TP53,TP530.40083333 Polycystic(Kidney,(Autosomal(Dominant KIRC 1 0.008828178 TSC2(@>(mtor(signaling(pathway(PTEN,PIK3CA)0.77024073 1 1 Spinocerebellar(Ataxia KIRC 1 1 0.010076664 JPH3,ATXN2,TTBK2,KCNC3,SYT14,PPP2R2B(@>(TSPAN19(2.27e@02);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(EPHA6(6.72e@04);JPH3,CACNA1A,ATXN2,SPTBN2,TTBK2,KCNC3,ATXN10,PRKCG,FGF14,SYT14,PPP2R2B(@>(GPRIN1(1.30e@02);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(CSMD1(3.20e@03);JPH3,ZNF592,CACNA1A,SYNE1,ATM,SPTBN2,TTBK2,KCNC3,ITPR1,FGF14,SETX,SYT14,PPP2R2B,ATXN1(@>(FAM200A(4.51e@03);JPH3,ZNF592,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ITPR1,PRKCG,FGF14,SETX,PPP2R2B,ATXN1(@>(QKI(3.76e@03);ZNF592,ATXN7,TDP1,SYNE1,ATM,TTBK2,TBP,ITPR1,SETX,ATXN1(@>(PIK3CA(7.15e@03);JPH3,CACNA1A,ATXN2,SPTBN2,TTBK2,KCNC3,PRKCG,FGF14,SYT14,PPP2R2B(@>(NEFH(3.84e@03);JPH3,ZNF592,CACNA1A,ATXN2,PDYN,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,AFG3L2,SYT14,PPP2R2B(@>(PARK2(2.54e@03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(PTPRN2(3.69e@03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,A1 0.63027586 Hypopituitarism KIRC 1 0.004667886 GH1(@>(regulation(of(eif@4e(and(p70s6(kinase(PIK3CA,PTEN);GH1(@>(mtor(signaling(pathway(PTEN,PIK3CA);BTK(@>(BCR(signaling(pathway(PIK3CA,PTEN)1 1 1 Combined(Heart(and(Skeletal(Defects KIRC 1 0.098019822 EP300(@>(hypoxia@inducible(factor(in(the(cardivascular(system(VHL);EP300,CREBBP(@>(HIF@2@alpha(transcription(factor(network(VHL,TCEB1);EP300,CREBBP(@>(il@7(signal(transduction(PIK3CA);CREBBP(@>(Signaling(events(mediated(by(Stem(cell(factor(receptor((c@Kit)(PIK3CA,PTEN)0.667858981 1 1 Tuberous(Sclerosis KIRC 1 0.008738373 TSC2,TSC1(@>(mtor(signaling(pathway(PTEN,PIK3CA)0.455452649 1 1 Severe(Combined(Immunodeficiency KIRC 1 1 0.015958615 ZAP70,IL2RG,CIITA,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(VHL(1.20e@03);CIITA,RFXAP,DCLRE1C,RAG2,RAG1(@>(PBRM1(2.88e@02)1 1 Specified(Hamartoses KIRC 0.03 VHL,PTEN 9.56E@72 PTEN(@>(skeletal(muscle(hypertrophy(is(regulated(via(akt@mtor(pathway(PIK3CA,PTEN);PTEN(@>(regulation(of(eif@4e(and(p70s6(kinase(PIK3CA,PTEN);VHL(@>(vegf(hypoxia(and(angiogenesis(VHL,PIK3CA);PTEN(@>(mtor(signaling(pathway(PTEN,PIK3CA);VHL(@>(hypoxia@inducible(factor(in(the(cardivascular(system(VHL);PTEN(@>(BCR(signaling(pathway(PIK3CA,PTEN);VHL(@>(HIF@2@alpha(transcription(factor(network(VHL,TCEB1);VHL(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(VHL,TCEB1,CDKN2A);PTEN(@>(pten(dependent(cell(cycle(arrest(and(apoptosis(PTEN,PIK3CA);PTEN(@>(Negative(regulation(of(the(PI3K/AKT(network(PTEN);PTEN(@>(Signaling(events(mediated(by(Stem(cell(factor(receptor((c@Kit)(PIK3CA,PTEN);PTEN(@>(CXCR4@mediated(signaling(events(PIK3CA,PTEN)0.40794741 0 PTEN(@>(PIK3CA;SDHB(@>(CDKN2A,UQCRFS1,UQCRFS1,TCEB11 Chronic(Granulomatous(Disease KIRP 1 1 0.087493235 NCF4(@>(SYTL3(3.78e@02);NCF2,NCF4,CYBB,CYBA(@>(MARCH1(2.06e@02);NCF2,NCF4,CYBB,CYBA(@>(SOD2(3.13e@02);NCF2,NCF4,CYBB,CYBA(@>(KLHL2(2.56e@02);NCF2,CYBB,CYBA(@>(PID1(1.85e@02);NCF2,CYBB,CYBA(@>(SNX9(1.69e@02);NCF2,NCF4,CYBB,CYBA(@>(HRH2(3.13e@02);NCF2,NCF4(@>(CEACAM8(2.87e@02);NCF2,NCF4,CYBB,CYBA(@>(TAGAP(2.62e@02);NCF2,NCF4(@>(FBXL13(2.99e@02);NCF2,NCF4,CYBB(@>(PIK3CB(2.68e@02);NCF2,NCF4,CYBB,CYBA(@>(IGF2R(2.31e@02);NCF2,NCF4,CYBB,CYBA(@>(FNIP2(1.64e@02);NCF2,NCF4,CYBB,CYBA(@>(DOT1L(1.53e@02);NCF2,NCF4,CYBB,CYBA(@>(WTAP(1.85e@02);NCF2,NCF4,CYBB,CYBA(@>(UIMC1(1.95e@02);NCF2,CYBB(@>(NAF1(2.69e@02)1 1 Spinocerebellar(Ataxia KIRP 1 1 0.020656922 JPH3,ZNF592,CACNA1A,ATXN2,SYNE1,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(ZDHHC14(3.03e@03);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(CPE(3.40e@03);JPH3,ZNF592,ATM,TTBK2,KCNC3,ITPR1,SETX,PPP2R2B(@>(MARCH1(3.33e@02);ZNF592,CACNA1A,TTBK2,TBP,KCNC3,ITPR1,FGF14,AFG3L2(@>(PPARGC1B(7.59e@03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(RAPGEF2(2.75e@03);SPTBN2,KCNC3,SYT14(@>(RXFP1(4.50e@02);JPH3,ZNF592,ATXN7,ATXN2,TDP1,SYNE1,ATM,TTBK2,TBP,KCNC3,ITPR1,FGF14,AFG3L2,PPP2R2B(@>(MAP3K4(4.34e@03);JPH3,ZNF592,CACNA1A,ATXN2,PDYN,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,AFG3L2,SYT14,PPP2R2B(@>(PARK2(2.75e@03);JPH3,ZNF592,CACNA1A,ATXN2,PDYN,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SETX,SYT14,PPP2R2B,ATXN1(@>(KIAA0825(2.35e@03);JPH3,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,ATXN10,PRKCG,FGF14,SYT14,PPP2R2B(@>(PCDH11X(1.81e@03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,I1 1 Neurofibromatosis KIRP 0.47 NF2 1 0.720639739 0 NF1(@>(NF2,CDKN2A 1 Severe(Combined(Immunodeficiency KIRP 1 1 0.003470279 ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(SYTL3(6.34e@03);ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D(@>(C6orf99(3.25e@03);IL2RG,JAK3,PTPRC,DCLRE1C(@>(HRH2(3.70e@02);ZAP70,IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,IL7R,DCLRE1C(@>(TAGAP(1.66e@02);IL2RG,ADA,PNP,PTPRC(@>(WTAP(3.67e@02);ZAP70,IL2RG,JAK3,PTPRC,IL7R,CD3D,DCLRE1C(@>(BCL11B(1.22e@03);IL2RG,RFXAP,RFX5,AK2,CD3D,DCLRE1C(@>(VRK1(2.43e@04);IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,DCLRE1C(@>(UIMC1(6.16e@03);AK2,IL7R(@>(RPL22(1.82e@04);ZAP70,IL2RG,JAK3,PTPRC,IL7R,CD3D,DCLRE1C(@>(IL32(4.78e@04);ADA,PNP(@>(NAF1(2.66e@02);NHEJ1,RFXANK,AK2(@>(TREX2(5.68e@03)1 1 Lipoprotein(Deficiencies KIRP 1 1 0.052743895 MTTP,APOB,LCAT,SAR1B,APOA1(@>(ETFDH(6.51e@03);APOB,LCAT,APOA1(@>(LPA(5.65e@03);ABCA1(@>(PID1(3.54e@02);MTTP,APOB,APOA1(@>(CEACAM1(2.64e@02);MTTP,APOB,LCAT,SAR1B,APOA1(@>(SLC22A1(6.51e@03);(@>(SLC22A3(2.64e@02);MTTP(@>(KCNK5(4.44e@02)1 1 Disorders(of(Urea(Cycle(Metabolism KIRP 1 1 0.088531104 ASS1,NAGS,ARG1,ASL,CPS1(@>(ETFDH(1.60e@02);ASS1,ARG1,CPS1(@>(LPA(1.67e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(SLC22A1(1.67e@02);NAGS(@>(MSMO1(1.63e@02)1 1 Chronic(Granulomatous(Disease LGG 1 0.438518035 0.076994822 (@>(CRLF2(2.65e@02);CYBA(@>(IRF4(2.99e@02);NCF2,NCF4,CYBB,CYBA(@>(DUSP22(2.36e@02);CYBA(@>(TWF2(1.86e@02);NCF2,NCF4,CYBB,CYBA(@>(ADAM8(1.08e@02);NCF2,NCF4,CYBB,CYBA(@>(PTEN(1.20e@02);NCF2,NCF4,CYBB,CYBA(@>(ATP9B(1.09e@02);NCF2,NCF4,CYBB,CYBA(@>(NFATC1(1.29e@02);NCF2,NCF4,CYBB,CYBA(@>(CSF2RA(1.43e@02);NCF2,NCF4,CYBB,CYBA(@>(METRNL(1.91e@02);NCF2,NCF4,CYBB,CYBA(@>(LRRK2(1.06e@02);NCF2,NCF4,CYBA(@>(B3GNTL1(1.22e@02);NCF2,NCF4,CYBB,CYBA(@>(KCNQ1(1.10e@02);NCF2,NCF4,CYBB,CYBA(@>(GLYCTK(1.33e@02);NCF2,NCF4(@>(PHF8(4.49e@02);NCF2,NCF4,CYBA(@>(MEF2D(1.52e@02);NCF2,CYBB,CYBA(@>(KMO(1.10e@02);NCF2,NCF4,CYBB,CYBA(@>(CTDP1(1.08e@02);NCF2,NCF4,CYBB,CYBA(@>(TNFSF13B(4.32e@02);NCF2,NCF4,CYBA(@>(AGAP2(2.75e@02);NCF2,NCF4,CYBB,CYBA(@>(IL3RA(1.77e@02);NCF2,NCF4,CYBB,CYBA(@>(PQLC1(1.26e@02);NCF2,NCF4(@>(IRS2(1.96e@02);NCF2,NCF4,CYBB,CYBA(@>(LPAR6(2.36e@02);NCF4,CYBB,CYBA(@>(STAB1(2.69e@02);NCF2,NCF4,CYBB,CYBA(@>(RB1(3.25e@02);CYBA(@>(HEATR3(3.40e@02);NCF2(@>(P2RY8(1.19e@02);NCF2,NCF4,CYBA(@>(RASA3(1.88e@02);NCF2,NCF4,CYBB,CYBA(1 1 Congenital(Ichthyosis LGG 1 0.297294297 0.009387698 ALOX12B,ALOXE3,CSTA,TGM1(@>(PARD6G(3.97e@02);(@>(SLC25A6(1.39e@03);CSTA,TGM1(@>(CYP27B1(4.66e@02);ALOX12B,ALDH3A2,ALOXE3,SPINK5,CSTA,KRT2,ABCA12,TGM1(@>(ZNF750(6.06e@04);ALOX12B,ALOXE3,SPINK5,CSTA,ABCA12,TGM1(@>(MPZL2(2.34e@03);ALOX12B,ALDH3A2,ALOXE3,SPINK5,CSTA,ABCA12,TGM1(@>(KLF5(1.39e@03);CSTA,NIPAL4,LIPN,ABHD5(@>(NOTCH1(1.90e@02)1 1 Inherited(Adrenogenital(Disorders LGG 1 0.709981324 0.078663977 HSD3B2,POR,CYP17A1,CYP21A2(@>(ITIH1(1.15e@02);HSD3B2,POR,CYP17A1,CYP21A2(@>(ITIH3(2.20e@02);HSD3B2,POR,CYP17A1,CYP21A2(@>(ITIH4(2.20e@02);HSD3B2,CYP17A1,CYP21A2(@>(ARSE(2.53e@02)1 1 Pervasive,(Specified(Congenital(Anomalies LGG 1 FGD1 4.83E@30 HRAS,RAF1,SOS1(@>(t(cell(receptor(signaling(pathway(NFATC1,CD3E,PIK3R1,PIK3CA);FGD1(@>(Regulation(of(CDC42(activity(FARP2,FGD1);HRAS,SOS1(@>(nfat(and(hypertrophy(of(the(heart((NFATC1,PIK3R1,PIK3CA);HRAS,NRAS,SOS1,KRAS(@>(Trk(receptor(signaling(mediated(by(PI3K(and(PLC@gamma(PIK3CA,AGAP2,PIK3R1);HRAS,RAF1,SOS1(@>(pdgf(signaling(pathway(PIK3CA,PDGFRA,PIK3R1);PTCH1(@>(Signaling(events(mediated(by(the(Hedgehog(family(PIK3CA,PIK3R1);SOS1(@>(PDGFR@alpha(signaling(pathway(PIK3CA,PDGFRA,PIK3R1);HRAS,SOS1(@>(vegf(hypoxia(and(angiogenesis(PIK3CA,PIK3R1);PTPN11,SOS1(@>(IL2(signaling(events(mediated(by(STAT5(PIK3CA,PIK3R1);PTPN11(@>(VEGFR1(specific(signals(PIK3CA,PIK3R1);HRAS,RAF1(@>(Nongenotropic(Androgen(signaling(PIK3CA,PIK3R1);HRAS,RAF1,SOS1(@>(tpo(signaling(pathway(PIK3CA,PIK3R1);HRAS,RAF1,SOS1(@>(multiple(antiapoptotic(pathways(from(igf@1r(signaling(lead(to(bad(phosphorylation(PIK3CA,PIK3R1);HRAS,RAF1,PTPN11,SOS1(@>(IGF1(pathway(PIK3CA,IRS2,PIK3R1);HRAS,RAF1,SOS1(@>(inhibition(of(cellular(proliferation(by(gleevec(P0.218652765 0.73112 0.26455 Diamond@Blackfan(Anemia LGG 1 1 0.020326331 RPS26,RPS24,RPL5,RPS7(@>(BORA(1.65e@03);RPS24,RPL5,RPS7(@>(GNL3(5.85e@03);RPS26,RPS19,RPS10,RPL35A,RPS7(@>(PTBP1(5.88e@03);RPS26(@>(CDKN2A(1.01e@02);RPS26,RPS19,RPS10(@>(TWF2(3.53e@02);RPL5(@>(TLR9(4.05e@02);(@>(VENTX(8.75e@03);RPS26,RPS19,RPS7(@>(TP53(1.87e@03);(@>(UTF1(9.49e@03);RPS26,RPS7(@>(METTL1(2.67e@02);RPS26(@>(CDK4(2.75e@02);RPS26,RPS19,RPS10,RPS7(@>(SPCS1(6.27e@03);RPS26,RPS19,RPS10,RPL35A,RPS7(@>(TXNL4A(3.42e@02);RPS26,RPS7(@>(HEATR3(9.43e@03);(@>(IL32(2.72e@02);RPS26,RPS7(@>(MZT1(1.39e@02)1 1 Inherited(Anomalies(of(the(Skin LGG 1 0.013851661 ATP2A2(@>(nfat(and(hypertrophy(of(the(heart((NFATC1,PIK3R1,PIK3CA);TERT(@>(telomeres(telomerase(cellular(aging(and(immortality(TP53,RB1);TERT(@>(role(of(nicotinic(acetylcholine(receptors(in(the(regulation(of(apoptosis(PIK3CA,PIK3R1)0.112939761 1 1 Spinocerebellar(Ataxia LGG 1 1 0.020326331 ATXN7,TDP1,SYNE1,ATM,TTBK2,TBP,ITPR1,SETX(@>(PIBF1(2.73e@02);(@>(PPP2R3B(2.73e@02);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,ATM,SPTBN2,TTBK2,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(NISCH(4.60e@03);JPH3,ATXN2,SYNE1,SPTBN2,TTBK2,KCNC3,PRKCG,SYT14,PPP2R2B(@>(IFLTD1(1.08e@02);JPH3,ZNF592,CACNA1A,ATXN7,ATXN2,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SETX,SYT14,PPP2R2B,ATXN1(@>(ATRX(1.74e@03);POLG,ATXN7,TDP1,ATM,TTBK2,TBP,ITPR1(@>(MARCH9(9.72e@03);ZNF592,ATXN7,TDP1,SYNE1,ATM,TTBK2,TBP,ITPR1,SETX,ATXN1(@>(PIK3CA(1.35e@02);JPH3,ZNF592,ATXN2,SPTBN2,TTBK2,KCNC3,ITPR1,PRKCG,FGF14,PPP2R2B(@>(PIK3C2B(1.06e@02);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(PALM(3.34e@02);APTX,ATXN7,TDP1,ATM,TBP,NOP56,SETX,C10orf2(@>(THOC1(3.47e@02);ZNF592,POLG,ATXN7,TDP1,SYNE1,ATM,TTBK2,TBP,ITPR1,SETX,ATXN1(@>(MDM4(2.02e@02);TDP1,TBP,ITPR1,NOP56,SETX,C10orf2,ATXN1(@>(FUBP1(2.67e@02);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(IGSF9B(1.67e@03);JPH3,APTX,1 1 Hypopituitarism LGG 1 0.001321061 GH1(@>(regulation(of(eif@4e(and(p70s6(kinase(PIK3CA,PIK3R1,PTEN);GLI2(@>(Signaling(events(mediated(by(the(Hedgehog(family(PIK3CA,PIK3R1);GH1(@>(mtor(signaling(pathway(PTEN,PIK3R1,PIK3CA);BTK(@>(BCR(signaling(pathway(NFATC1,PIK3CA,PIK3R1,PTEN);GH1(@>(growth(hormone(signaling(pathway(PIK3CA,PIK3R1);BTK(@>(EPO(signaling(pathway(PIK3R1,IRS2);GH1(@>(akt(signaling(pathway(PIK3CA,PIK3R1)0.15528125 1 1 Combined(Heart(and(Skeletal(Defects LGG 1 0.002464787 CREBBP(@>(nfat(and(hypertrophy(of(the(heart((NFATC1,PIK3R1,PIK3CA);EP300,CREBBP(@>(p53(pathway(TP53,CDKN2A,MDM4);EP300(@>(role(of(mef2d(in(t@cell(apoptosis(NFATC1,CD3E,MEF2D);CREBBP(@>(regulation(of(transcriptional(activity(by(pml(TP53,RB1);EP300,CREBBP(@>(E2F(transcription(factor(network(CDKN2C,CDKN2A,RB1);EP300,CREBBP(@>(il@7(signal(transduction(PIK3CA,PIK3R1);CREBBP(@>(Signaling(events(mediated(by(Stem(cell(factor(receptor((c@Kit)(PTEN,PIK3CA,PIK3R1,PIK3C2B);EP300(@>(ATF@2(transcription(factor(network(PDGFRA,CDK4,RB1);EP300,CREBBP(@>(FOXM1(transcription(factor(network(CDKN2A,CDK4,RB1)0.651597416 0.55437288 0.63027586 Hereditary(Sensory(Neuropathy LGG 1 1 0.099201187 (@>(DHRSX(1.98e@02);(@>(SLC25A6(4.63e@02);(@>(ASMTL(1.97e@02);LMNA(@>(CD99(4.74e@02);SBF2,PRX,LMNA,HSPB8,MTMR2,SH3TC2,PMP22,KIF1B,NEFL,INF2(@>(SASH1(1.90e@02);(@>(ZBED1(1.98e@02)1 1 Severe(Combined(Immunodeficiency LGG 1 2.73E@09 CD3D,PTPRC,ZAP70(@>(t(cell(receptor(signaling(pathway(NFATC1,CD3E,PIK3R1,PIK3CA);CD3D(@>(Downstream(TCR(signaling(PTEN,CD3E);JAK3,IL2RG(@>(IL2(signaling(events(mediated(by(STAT5(PIK3CA,PIK3R1);PTPRC(@>(BCR(signaling(pathway(NFATC1,PIK3CA,PIK3R1,PTEN);CD3D,PTPRC,ZAP70(@>(role(of(mef2d(in(t@cell(apoptosis(NFATC1,CD3E,MEF2D);CD3D(@>(the(co@stimulatory(signal(during(t@cell(activation(PIK3CA,CD3E,PIK3R1);JAK3,IL2RG(@>(il@7(signal(transduction(PIK3CA,PIK3R1);JAK3,IL2RG(@>(IL4@mediated(signaling(events(PIK3CA,PIK3R1,IRS2,IRF4);JAK3,IL2RG(@>(IL2@mediated(signaling(events(PIK3CA,IRS2,PIK3R1);CD3D,PTPRC(@>(CXCR4@mediated(signaling(events(PIK3CA,CD3E,PIK3R1,PTEN)0.001993726 ZAP70,IL2RG,JAK3,RFXAP,IL7R,CD3D,DCLRE1C(@>(PIBF1(2.80e@02);IL2RG,CIITA,RFX5,DCLRE1C(@>(IRF4(1.12e@03);ZAP70,IL2RG,JAK3,RFXAP,RFX5,IL7R,CD3D,DCLRE1C(@>(MARCH9(1.01e@02);IL2RG,RFXAP,RFX5,AK2,CD3D,DCLRE1C(@>(BORA(1.01e@04);ZAP70,IL2RG,RFXAP,RFX5,AK2,IL7R,CD3D,DCLRE1C(@>(THOC1(6.29e@03);ZAP70,IL2RG,CIITA,JAK3,RFXAP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C,RAG2,RAG1(@>(MDM4(3.01e@03);IL2RG,JAK3,PNP,RFX5,PTPRC(@>(ADAM8(4.10e@02);CIITA,RFX5,DCLRE1C(@>(TLR9(8.90e@05);ZAP70,IL2RG,JAK3,PTPRC,IL7R,CD3D,DCLRE1C,RAG2,RAG1(@>(CD3E(3.13e@04);ZAP70,IL2RG,CIITA,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(ATP9B(1.20e@02);IL2RG,JAK3,ADA,PNP,RFX5,PTPRC(@>(NFATC1(1.78e@02);IL2RG,RFXAP,AK2,DCLRE1C(@>(FUBP1(1.00e@02);ZAP70,IL2RG,JAK3,PTPRC,IL7R,CD3D,DCLRE1C(@>(PPM1M(4.94e@03);IL2RG,JAK3,PTPRC,AK2,DCLRE1C(@>(B3GNTL1(1.63e@04);RFXAP,RFX5,AK2,DCLRE1C(@>(NCAPD3(2.56e@03);ZAP70,IL2RG,JAK3,PNP,PTPRC,IL7R,CD3D,DCLRE1C(@>(GLYCTK(1.19e@02);ZAP70,IL2RG,JAK3,RFXAP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(PHF8(1.18e@02);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD30.28232609 1 Specified(Hamartoses LGG 0.66 PTEN 1.74E@07 PTEN(@>(skeletal(muscle(hypertrophy(is(regulated(via(akt@mtor(pathway(PIK3CA,PIK3R1,PTEN);PTEN(@>(regulation(of(eif@4e(and(p70s6(kinase(PIK3CA,PIK3R1,PTEN);VHL(@>(vegf(hypoxia(and(angiogenesis(PIK3CA,PIK3R1);PTEN(@>(Downstream(TCR(signaling(PTEN,CD3E);PTEN(@>(mtor(signaling(pathway(PTEN,PIK3R1,PIK3CA);PTEN(@>(BCR(signaling(pathway(NFATC1,PIK3CA,PIK3R1,PTEN);VHL(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53);PTEN(@>(pten(dependent(cell(cycle(arrest(and(apoptosis(PTEN,PIK3R1,PIK3CA);PTEN(@>(Signaling(events(mediated(by(Stem(cell(factor(receptor((c@Kit)(PTEN,PIK3CA,PIK3R1,PIK3C2B);PTEN(@>(CXCR4@mediated(signaling(events(PIK3CA,CD3E,PIK3R1,PTEN)0.613735058 0.20252632 PTEN(@>(PIK3CA;STK11(@>(CDKN2A1 Holoprosencephaly LGG 1 1 0.043065539 TDGF1,NODAL,FOXH1,GLI2(@>(VENTX(4.40e@03);TDGF1,NODAL,FOXH1,GLI2(@>(UTF1(1.43e@02);TDGF1,FOXH1,ZIC2(@>(PLCXD1(1.13e@02);(@>(MID1(1.43e@02)1 1 Li(Fraumeni(and(Related(Syndromes LGG 0.03 CDKN2A,TP53 4.44E@10 TP53(@>(chaperones(modulate(interferon(signaling(pathway(TP53,RB1);TP53,CDKN2A,CHEK2(@>(p53(pathway(TP53,CDKN2A,MDM4);TP53(@>(estrogen(responsive(protein(efp(controls(cell(cycle(and(breast(tumors(growth(CDK4,TP53);CDKN2A,TP53(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53);TP53(@>(telomeres(telomerase(cellular(aging(and(immortality(TP53,RB1);TP53(@>(btg(family(proteins(and(cell(cycle(regulation(TP53,RB1);TP53(@>(Transcriptional((activation(of((cell(cycle(inhibitor(p21(TP53);TP53(@>(p53(signaling(pathway(CDK4,TP53,RB1);TP53(@>(regulation(of(transcriptional(activity(by(pml(TP53,RB1);TP53(@>(rb(tumor(suppressor/checkpoint(signaling(in(response(to(dna(damage(CDK4,TP53,RB1);CDKN2A(@>(E2F(transcription(factor(network(CDKN2C,CDKN2A,RB1);TP53(@>(p75(NTR)@mediated(signaling(PIK3CA,TP53,PIK3R1);CDKN2A,CHEK2(@>(FOXM1(transcription(factor(network(CDKN2A,CDK4,RB1)0.667858981 0 CDKN2A(@>(CDKN2C;CHEK2(@>(CDK4;TP53(@>(PTEN,TP53,GNL30.31565625 Lipoprotein(Deficiencies LGG 1 0.563318142 0.041865488 APOB,SAR1B(@>(IDH1(1.07e@02);MTTP,APOB,LCAT,APOA1,ABCA1(@>(GLYCTK(8.69e@03);MTTP,APOB,LCAT,SAR1B,APOA1(@>(ITIH1(7.83e@03);MTTP,APOB,LCAT,SAR1B,APOA1(@>(ITIH3(7.83e@03);MTTP,APOB,LCAT,SAR1B,APOA1(@>(ITIH4(4.19e@03);MTTP,APOB,LCAT,APOA1(@>(ARSD(6.80e@03);MTTP,APOB,LCAT,SAR1B,APOA1(@>(ARSE(6.11e@03)1 1 Disorders(of(Urea(Cycle(Metabolism LGG 1 0.01382358 ARG1(@>(IL4@mediated(signaling(events(PIK3CA,PIK3R1,IRS2,IRF4);ARG1(@>(ATF@2(transcription(factor(network(PDGFRA,CDK4,RB1)0.066215584 ASS1,NAGS,ASL,CPS1(@>(HSBP1L1(7.80e@03);NAGS,ARG1,ASL,CPS1(@>(GLYCTK(3.87e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(ITIH1(2.55e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(ITIH3(1.30e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(ITIH4(2.55e@02);NAGS,ARG1,ASL(@>(IDH2(2.02e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(ARSD(8.13e@03);ASS1,NAGS,ARG1,ASL,CPS1(@>(ARSE(8.74e@03)1 1 Retinitis(Pigmentosa LGG 1 0.611253215 1.45E@10 SNRNP200,ZNF513,CRB1,PDE6B,RP9,FAM161A(@>(MARCH9(6.27e@03);(@>(DHRSX(2.24e@03);CRX,FSCN2,RDH12,PRPH2,CNGB1,EYS,CRB1,CERKL,TULP1,ROM1,PRCD,IMPG2,C2orf71,RBP3(@>(ASMT(5.31e@09);TTC8,KLHL7,EYS,SPATA7,CRB1,MERTK,PDE6B,CERKL,PRCD,IMPG2,C2orf71,RBP3,FAM161A(@>(PCDHAC2(9.96e@03);IMPDH1(@>(CYP27B1(1.88e@02);(@>(ZBED1(5.82e@03);CRX,FSCN2,RDH12,SNRNP200,PRPH2,RHO,CNGB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,USH2A,PROM1,CNGA1,RP1,PRCD,RPE65,SAG,RBP3,FAM161A,ABCA4(@>(GLB1L2(2.11e@12);CRX,FSCN2,PRPH2,CNGB1,EYS,CRB1,CERKL,TULP1,PRCD,IMPG2,C2orf71,RBP3(@>(MYO16(2.31e@05);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,RP1,PRCD,RPE65,SAG,RBP3,ABCA4(@>(KCNJ13(4.64e@12);FAM161A(@>(SHOX(3.26e@10);SPATA7,CRB1,PDE6B(@>(DRD1(1.42e@02)0.57574242 1 Osteogenesis(Imperfecta LGG 1 0.521588572 0.076277809 (@>(SLC25A6(2.01e@02);CRTAP,LEPRE1(@>(TWF2(1.35e@02);CRTAP,COL1A2(@>(METRNL(2.47e@02);(@>(DCP1B(2.51e@02);COL1A2,COL1A1,LEPRE1(@>(CDC16(1.25e@02);COL1A2,LEPRE1(@>(JAM3(4.12e@02);CRTAP,COL1A2,COL1A1,LEPRE1(@>(CD99(4.89e@02);(@>(TSPAN31(1.60e@02);CRTAP,COL1A2(@>(NAB2(1.38e@02);CRTAP,COL1A2,COL1A1,LEPRE1(@>(PDGFRA(3.31e@02);CRTAP(@>(SMIM4(2.47e@02);CRTAP,COL1A2,COL1A1,LEPRE1(@>(SASH1(1.40e@02);CRTAP,COL1A2,COL1A1,LEPRE1(@>(FGD1(1.32e@02);COL1A2(@>(XG(2.25e@02);CRTAP,COL1A2,COL1A1,LEPRE1(@>(NT5DC2(1.17e@02);(@>(CDK4(2.15e@02);LEPRE1(@>(NOX4(1.21e@02);CRTAP,LEPRE1(@>(SPCS1(1.16e@02);CRTAP,COL1A2,COL1A1,LEPRE1(@>(TEAD3(1.91e@02);CRTAP,LEPRE1(@>(TXNL4A(4.07e@02);(@>(SHOX(1.17e@02);COL1A2,COL1A1,LEPRE1(@>(MXRA5(2.58e@02);CRTAP,COL1A2,COL1A1(@>(PRCP(1.20e@02)1 1 Anophthalmos/Micropthalmos LGG 1 1 0.070668747 VSX2,MFRP,RAX(@>(GLB1L2(8.83e@03)1 1 Chronic(Granulomatous(Disease LUAD 1 0.384190056 0.078679242 NCF4,CYBA(@>(CCND3(3.54e@02);NCF2,NCF4,CYBA(@>(ZGPAT(1.74e@02);NCF2,NCF4,CYBB,CYBA(@>(RIT1(1.48e@02);NCF2,NCF4,CYBB,CYBA(@>(ITGAX(1.79e@02);NCF2,NCF4,CYBB,CYBA(@>(DNAJC5(1.82e@02);NCF2,NCF4,CYBB,CYBA(@>(PTGER4(1.60e@02);NCF2,NCF4,CYBB,CYBA(@>(SIRPB1(1.30e@02);NCF2,NCF4,CYBB,CYBA(@>(TNFSF13B(4.88e@02);NCF2,NCF4,CYBB,CYBA(@>(PQLC1(1.33e@02);NCF2,NCF4,CYBB,CYBA(@>(TBL1X(1.72e@02);NCF4,CYBA(@>(ARHGEF6(2.95e@02);NCF2,NCF4,CYBB,CYBA(@>(GNG2(1.37e@02);NCF2,NCF4,CYBB,CYBA(@>(NFATC1(1.81e@02);NCF2,NCF4(@>(U2AF1(3.61e@02);NCF2,NCF4,CYBB,CYBA(@>(BTK(1.94e@02);NCF2,NCF4,CYBB,CYBA(@>(SAMSN1(1.35e@02);NCF2,NCF4,CYBB,CYBA(@>(LPAR6(2.19e@02);NCF2,NCF4(@>(IL18RAP(1.83e@02);NCF2,NCF4,CYBB,CYBA(@>(RB1(3.72e@02);NCF2,NCF4,CYBB(@>(ADNP2(3.36e@02);NCF2,NCF4,CYBB,CYBA(@>(MFSD7(3.07e@02);NCF2,NCF4,CYBB,CYBA(@>(PMAIP1(1.26e@02);NCF4,CYBB,CYBA(@>(TBX21(1.81e@02);NCF2,NCF4,CYBB,CYBA(@>(ANKRD44(1.82e@02);NCF2,NCF4,CYBB,CYBA(@>(AQP9(1.28e@02);NCF2,NCF4,CYBB,CYBA(@>(PPM1F(1.53e@02);NCF2,NCF4,CYBB,CYBA(@>(CTDP1(1.37e@02);NCF2,NCF4,CYBB,CYB1 1 Congenital(Ichthyosis LUAD 1 1 0.020062495 ALOX12B,SPINK5,CSTA,TGM1(@>(SPRR3(8.05e@03);ALOX12B(@>(KRT28(3.04e@03);(@>(FLG(1.68e@03);ALOX12B,SPINK5,CSTA,ABCA12,TGM1(@>(SERPINB13(1.68e@03);ALOX12B,SPINK5,KRT2,ABCA12(@>(POF1B(8.05e@03)1 1 Disorders(of(Phosphorous(Metabolism LUAD 1 CYP27B1 0.003900738 FGF23(@>(Syndecan@2@mediated(signaling(events(HRAS,LAMA3,FGF19,NF1);CYP27B1(@>(Vitamin(D((calciferol)(metabolism(GC,CYP27B1);FGF23(@>(Syndecan@1@mediated(signaling(events(COL11A1,COL5A2,MET,FGF19,COL5A1,COL3A1)1 1 1 Diamond@Blackfan(Anemia LUAD 1 1 0.006070507 RPS26,RPS19,RPS10,RPS7(@>(BYSL(4.40e@02);RPL5,RPS10,RPL11(@>(CCND3(3.53e@02);(@>(UTF1(8.74e@03);RPS24,RPL5(@>(ALG10(3.42e@02);RPS26,RPS19,RPS7(@>(GEMIN4(2.58e@02);RPS26(@>(CDKN2A(9.01e@03);RPS26,RPL35A,RPS7(@>(U2AF1(4.63e@04);RPS26,RPS7(@>(FANCD2(1.47e@02);RPS26,RPS19,RPS10,RPL35A(@>(LAGE3(4.10e@02);(@>(CTCFL(1.14e@02);(@>(MDM2(4.11e@02);(@>(VENTX(7.84e@03);RPS26,RPS19,RPS7(@>(TP53(1.67e@03);RPS26,RPS19,RPS10,RPS7(@>(EIF4EBP1(1.12e@02);RPS26,RPS19,RPS7(@>(TFDP1(2.13e@03);RPS26,RPS19,RPS10,RPL35A,RPS7(@>(TXNL4A(3.14e@02);(@>(TERT(1.10e@02);RPS19,RPS7(@>(CNIH1(4.86e@04);RPL5,RPS7(@>(NRAS(1.31e@03);RPS26,RPS10,RPL35A(@>(RNMTL1(3.54e@04);RPS26,RPS7(@>(METTL1(2.41e@02);RPS26,RPS19,RPS10,RPS7(@>(HAX1(1.49e@02)1 1 Inherited(Anomalies(of(the(Skin LUAD 1 TERT 3.07E@07 TERT(@>(overview(of(telomerase(protein(component(gene(htert(transcriptional(regulation(TERT,TP53);TERT(@>(telomeres(telomerase(cellular(aging(and(immortality(TERT,TP53,RB1,KRAS);KRT1,TERT(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(CDKN2A,TBL1X,CCND1,APC,TERT,SMARCA4)0.086097896 KRT6A,KRT16(@>(CYP27B1(1.45e@02);WRAP53,TERC,DKC1,NHP2,NOP10(@>(GEMIN4(4.63e@02);TERC,NHP2,NOP10(@>(FAM58A(4.53e@02);KRT6A,TERC,NHP2,KRT16,NOP10(@>(HRAS(4.57e@02);TERC,NHP2,NOP10(@>(SLC10A3(4.68e@02);KRT6A,NHP2,KRT16,NOP10(@>(NXN(4.37e@02);(@>(AKR1B10(4.64e@02)1 1 Spinocerebellar(Ataxia LUAD 1 ATM 8.60E@07 PRKCG(@>(EGFR(Inhibitor(Pathway,(Pharmacodynamics(ERBB2,NRAS,HRAS,EGFR,KRAS);ATM(@>(apoptotic(signaling(in(response(to(dna(damage(ATM,TP53);ATM(@>(p53(pathway(CDKN2A,ATM,TP53,MDM2);ATM(@>(ATM(pathway(MDM2,ATM,FANCD2);ATM(@>(cell(cycle:(g2/m(checkpoint(MDM2,ATM,TP53,MYT1);ATM(@>(ATM(mediated(response(to(DNA(double@strand(break(ATM);ATM(@>(cdc25(and(chk1(regulatory(pathway(in(response(to(dna(damage(ATM,MYT1);ATM(@>(role(of(brca1(brca2(and(atr(in(cancer(susceptibility(ATM,TP53,FANCD2);ATM(@>(BARD1(signaling(events(ATM,TP53,FANCD2);ATM(@>(atm(signaling(pathway(MDM2,ATM,TP53);PRKCG(@>(IL8@(and(CXCR2@mediated(signaling(events(PLCB1,GNG2,HCK);ATM(@>(rb(tumor(suppressor/checkpoint(signaling(in(response(to(dna(damage(ATM,TP53,RB1,MYT1);PRKCG(@>(IL8@(and(CXCR1@mediated(signaling(events(PLCB1,GNG2,HCK);ATM(@>(Metformin(Pathway,(Pharmacodynamic(ATM,STK11,PRKAA1);ATM(@>(E2F(transcription(factor(network(TFDP1,CCND3,ATM,CDKN2A,RB1);ATM(@>(hypoxia(and(p53(in(the(cardiovascular(system(MDM2,ATM,TP53);TBP(@>(Glucocorticoid(rece0.001274155 SYT14(@>(SCG2(2.09e@02);APTX,ZNF592,ATXN2,TTBK2,TBP,KCNC3,ITPR1,NOP56,SETX,SYT14,C10orf2(@>(KIAA0907(2.58e@03);JPH3,TDP1,KCNC3,FGF14(@>(SAMD10(1.63e@02);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(DOC2B(3.57e@03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(C1orf173(3.57e@03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(TMEM132D(4.61e@03);JPH3,APTX,ZNF592,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,TBP,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(NF1(2.40e@03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(OPCML(2.50e@03);JPH3,ZNF592,SPTBN2,TTBK2,KCNC3,PRNP,PRKCG,FGF14,PPP2R2B,ATXN1(@>(DNAJC5(3.21e@02);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,TBP,KCNC3,PRNP,ITPR1,PRKCG,FGF14,SETX,SYT14,PPP2R2B(@>(TTC33(2.02e@03);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,TBP,KCNC3,ATXN10,ITPR1,PRKCG,FGF14,AFG3L2,SYT14,PPP2R2 Glucose@6@Phosphate(Dehydrogenase(DeficiencyLUAD 0.05 UBL4A,G6PD 0.034828736 NA(@>(NA(NA) 121 @1 1 1 Hypopituitarism LUAD 1 BTK 0.034828736 FGFR1(@>(Syndecan@2@mediated(signaling(events(HRAS,LAMA3,FGF19,NF1);BTK(@>(bcr(signaling(pathway(NFATC1,HRAS,BTK,PPP3CA);BTK(@>(EPO(signaling(pathway(HRAS,PTPN11,BTK);POU1F1(@>(Glucocorticoid(receptor(regulatory(network(NFATC1,TP53,TBX21,MDM2,SMARCA4,GATA3);GH1(@>(trefoil(factors(initiate(mucosal(healing(ERBB2,HRAS,EGFR);FGFR1(@>(Syndecan@1@mediated(signaling(events(COL11A1,COL5A2,MET,FGF19,COL5A1,COL3A1)1 0.2886 1 Combined(Heart(and(Skeletal(Defects LUAD 1 3.78E@08 EP300,CREBBP(@>(Direct(p53(effectors(BCL2L14,PMAIP1,TP53,EGFR,RB1,MET,APC,MDM2,SMARCA4,TFDP1);EP300,CREBBP(@>(p53(pathway(CDKN2A,ATM,TP53,MDM2);EP300(@>(cell(cycle:(g2/m(checkpoint(MDM2,ATM,TP53,MYT1);EP300(@>(Validated(transcriptional(targets(of(AP1(family(members(Fra1(and(Fra2(CCND1,NFATC1,LAMA3,CDKN2A);EP300(@>(Validated(transcriptional(targets(of(TAp63(isoforms(MDM2,CDKN2A,PMAIP1,SPATA18);EP300(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(CDKN2A,TBL1X,CCND1,APC,TERT,SMARCA4);CREBBP(@>(regulation(of(transcriptional(activity(by(pml(TP53,RB1);EP300,CREBBP(@>(E2F(transcription(factor(network(TFDP1,CCND3,ATM,CDKN2A,RB1);EP300(@>(p73(transcription(factor(network(JAG2,MDM2,NTRK1,RB1,WWOX);EP300(@>(hypoxia(and(p53(in(the(cardiovascular(system(MDM2,ATM,TP53);EP300,CREBBP(@>(Glucocorticoid(receptor(regulatory(network(NFATC1,TP53,TBX21,MDM2,SMARCA4,GATA3);EP300(@>(melanocyte(development(and(pigmentation(pathway(HRAS,KIT)0.71670305 1 1 Neurofibromatosis LUAD 0.66 NF1 0.001098828 NF1(@>(Regulation(of(Ras(family(activation(HRAS,NRAS,NF1,KRAS);NF1(@>(Syndecan@2@mediated(signaling(events(HRAS,LAMA3,FGF19,NF1);NF1(@>(chromatin(remodeling(by(hswi/snf(atp@dependent(complexes(ARID1A,SMARCA4,NF1)0.675271975 0.22904762 NF2(@>(CDKN2A 0.60125 Hereditary(Sensory(Neuropathy LUAD 1 NTRK1 1.95E@05 NTRK1(@>(Trk(receptor(signaling(mediated(by(PI3K(and(PLC@gamma(CCND1,HRAS,NRAS,NTRK1,KRAS);NTRK1(@>(ARMS@mediated(activation(NTRK1,BRAF);NDRG1(@>(Direct(p53(effectors(BCL2L14,PMAIP1,TP53,EGFR,RB1,MET,APC,MDM2,SMARCA4,TFDP1);EGR2(@>(Validated(transcriptional(targets(of(TAp63(isoforms(MDM2,CDKN2A,PMAIP1,SPATA18);NTRK1(@>(TRKA(activation(by(NGF(NTRK1);NTRK1(@>(NGF(signalling(via(TRKA(from(the(plasma(membrane(NTRK1);NTRK1(@>(Signalling(to(ERKs(NTRK1);NTRK1(@>(Signalling(to(STAT3(NTRK1);NTRK1(@>(trka(receptor(signaling(pathway(HRAS,NTRK1);RAB7A(@>(IL8@(and(CXCR2@mediated(signaling(events(PLCB1,GNG2,HCK);NTRK1(@>(Frs2@mediated(activation(NTRK1,BRAF);NTRK1(@>(Signalling(to(p38(via(RIT(and(RIN(NTRK1,BRAF);NTRK1(@>(p73(transcription(factor(network(JAG2,MDM2,NTRK1,RB1,WWOX);PMP22(@>(a6b1(and(a6b4(Integrin(signaling(ERBB2,MET,HRAS,LAMA3,EGFR)0.128961218 1 1 Severe(Combined(Immunodeficiency LUAD 1 0.009297242 DCLRE1C(@>(ATM(pathway(MDM2,ATM,FANCD2);ADA(@>(Validated(transcriptional(targets(of(TAp63(isoforms(MDM2,CDKN2A,PMAIP1,SPATA18);ADA(@>(p73(transcription(factor(network(JAG2,MDM2,NTRK1,RB1,WWOX);JAK3,IL2RG(@>(IL2@mediated(signaling(events(HRAS,NRAS,PTPN11,KRAS);JAK3(@>(il(6(signaling(pathway(PTPN11,HRAS);ADA(@>(Validated(transcriptional(targets(of(deltaNp63(isoforms(COL5A1,CDKN2A,ATM,MDM2)0.001684618 CIITA,RFX5,DCLRE1C(@>(C11orf35(2.10e@02);(@>(GATA3(2.09e@02);ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C,RAG2,RAG1(@>(CCND3(1.40e@04);ZAP70,IL2RG,JAK3,ADA,PTPRC,IL7R,CD3D(@>(ZGPAT(2.99e@03);ZAP70,IL2RG,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,CD3D(@>(PTGER4(1.27e@02);IL2RG,CIITA,JAK3,ADA,PNP,PTPRC(@>(TNFSF13B(4.44e@02);RFXANK(@>(RBM10(2.42e@03);ZAP70,IL2RG,RFXAP,RFX5,AK2,IL7R,CD3D,DCLRE1C(@>(ALG10(7.50e@04);IL2RG,RFX5,DCLRE1C(@>(CMTR2(2.45e@03);IL2RG,JAK3,PNP,RFX5,PTPRC,DCLRE1C(@>(AOAH(2.05e@02);IL2RG,JAK3,PTPRC(@>(TBL1X(2.39e@02);ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(ARHGEF6(9.58e@04);ZAP70,IL2RG,CIITA,JAK3,RFXAP,RFX5,PTPRC,CD3D,DCLRE1C(@>(ARID1A(5.42e@03);ZAP70,IL2RG,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,DCLRE1C(@>(GNG2(3.39e@02);IL2RG,JAK3,ADA,PNP,RFX5,PTPRC(@>(NFATC1(1.94e@02);ADA,PNP,AK2(@>(U2AF1(9.03e@04);ZAP70,IL2RG,JAK3,RFXAP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C,RAG2,RAG1(@>(ARID2(1.03e@02);IL2RG,CIITA,JAK3,RFX5,PTPRC,DCLRE1C(@>(BTK(3.08e@03);ZAP70,IL2RG,JAK3,PTPRC,IL7R,CD3D,DCLRE1C,RAG2,RAG1(@>(THEMIS(2.35e@04);RFXA1 1 Specified(Hamartoses LUAD 0.97 STK11 2.59E@05 VHL(@>(vegf(hypoxia(and(angiogenesis(HRAS,KDR,ARNT);PTEN(@>(Direct(p53(effectors(BCL2L14,PMAIP1,TP53,EGFR,RB1,MET,APC,MDM2,SMARCA4,TFDP1);VHL(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53,ARNT);STK11(@>(Metformin(Pathway,(Pharmacodynamic(ATM,STK11,PRKAA1)0.717635623 0.25727907 1 Li(Fraumeni(and(Related(Syndromes LUAD 0.09 CDKN2A,TP53 4.67E@28 TP53(@>(chaperones(modulate(interferon(signaling(pathway(TP53,RB1);TP53(@>(apoptotic(signaling(in(response(to(dna(damage(ATM,TP53);TP53(@>(Direct(p53(effectors(BCL2L14,PMAIP1,TP53,EGFR,RB1,MET,APC,MDM2,SMARCA4,TFDP1);TP53,CDKN2A,CHEK2(@>(p53(pathway(CDKN2A,ATM,TP53,MDM2);CHEK2(@>(ATM(pathway(MDM2,ATM,FANCD2);TP53,CHEK2(@>(cell(cycle:(g2/m(checkpoint(MDM2,ATM,TP53,MYT1);CDKN2A(@>(Validated(transcriptional(targets(of(AP1(family(members(Fra1(and(Fra2(CCND1,NFATC1,LAMA3,CDKN2A);CDKN2A(@>(Validated(transcriptional(targets(of(TAp63(isoforms(MDM2,CDKN2A,PMAIP1,SPATA18);TP53(@>(overview(of(telomerase(protein(component(gene(htert(transcriptional(regulation(TERT,TP53);TP53,CHEK2(@>(role(of(brca1(brca2(and(atr(in(cancer(susceptibility(ATM,TP53,FANCD2);CDKN2A,TP53(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53,ARNT);TP53(@>(telomeres(telomerase(cellular(aging(and(immortality(TERT,TP53,RB1,KRAS);CDKN2A(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(CDKN2A,TBL1X,CCND10.719639888 0 CDKN2A(@>(MDM2;CHEK2(@>(CCND3,ATM;TP53(@>(TP53,SMAD41 Genetic(Anomalies(of(Leukocytes LUAD 1 0.02688935 ITGB2(@>(Beta2(integrin(cell(surface(interactions(ITGAX,SPON2,FGB)0.471117151 1 1 Lipoprotein(Deficiencies LUAD 1 MTTP 1 0.097427692 APOB,LCAT,SAR1B,APOA1(@>(SLC26A1(2.83e@02);APOB,LCAT,SAR1B,APOA1(@>(PCK1(3.92e@02);APOB,LCAT,SAR1B,APOA1(@>(PROS1(1.82e@02);(@>(AKR1C2(2.10e@02);APOB,LCAT,SAR1B,APOA1(@>(EHHADH(1.83e@02);APOB,LCAT,SAR1B,APOA1(@>(BHMT(1.95e@02);APOB,LCAT,SAR1B,APOA1(@>(GBA3(1.93e@02);APOB,LCAT,SAR1B,APOA1(@>(ABCG5(2.40e@02);APOB,LCAT,SAR1B,APOA1(@>(MTTP(2.00e@02);(@>(CD5L(2.00e@02)1 1 Disorders(of(Urea(Cycle(Metabolism LUAD 1 1 0.069332922 ASS1,NAGS,ARG1,ASL,CPS1(@>(GC(1.50e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(SLC26A1(1.99e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(FGB(2.95e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(PROS1(1.02e@02);ASS1,NAGS,ASL,CPS1(@>(HSBP1L1(9.88e@03);ASS1(@>(AKR1C2(3.55e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(EHHADH(1.20e@02);NAGS,ARG1,ASL,CPS1(@>(SOWAHB(1.38e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(CYP4V2(3.55e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(GBA3(8.52e@03);ASS1,NAGS,ARG1,ASL,CPS1(@>(ABCG5(9.51e@03);ASS1,NAGS,ARG1,ASL,CPS1(@>(MTTP(9.88e@03);(@>(CD5L(9.22e@03)1 1 Retinitis(Pigmentosa LUAD 1 PDE6B 0.828420314 6.10E@15 RPGR,CRX,SNRNP200,CA4,EYS,CRB1,CERKL,PRPF3,TULP1,C2orf71,TOPORS,FAM161A(@>(KIAA0907(3.68e@02);KLHL7,SPATA7,CRB1,MERTK,CERKL,FAM161A(@>(DOC2B(3.41e@02);(@>(MUC16(2.78e@04);SNRNP200,CRB1,PRPF31,CERKL,IDH3B,PRPF8,FAM161A(@>(UCKL1(2.83e@03);CRX,FSCN2,RDH12,PRPH2,CNGB1,EYS,CRB1,CERKL,TULP1,ROM1,PRCD,IMPG2,C2orf71,RBP3(@>(GPR112(5.36e@07);KLHL7,SPATA7,CRB1,PRCD,IMPG2,C2orf71,FAM161A(@>(TMEM132D(3.43e@02);SNRNP200(@>(FZD10(1.68e@02);IMPDH1(@>(CYP27B1(1.33e@02);CRX,SNRNP200,KLHL7,EYS,SPATA7,CRB1,CERKL,USH2A,PRCD,IMPG2,C2orf71,RBP3,FAM161A(@>(GTF2I(4.98e@02);(@>(SLC22A6(1.68e@02);CNGA1(@>(ANKRD37(1.67e@02);CRX,FSCN2,PRPH2,CNGB1,KLHL7,EYS,SPATA7,CRB1,MERTK,CERKL,TULP1,ROM1,PRCD,IMPG2,C2orf71,RBP3,FAM161A(@>(RIMS2(3.67e@04);SPATA7,CRB1,PRPF3(@>(ITGB8(1.07e@02);RPGR,CA4,IMPDH1,PRPF3,BEST1,RP2,TOPORS,SEMA4A(@>(TBL1X(2.76e@02);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,EYS,CRB1,NRL,PDE6A,PDE6G,CERKL,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,PROM1,CNGA1,RP1,PRCD,IMPG2,C2orf71,RPE65,SAG,RBP3,FAM161A,ABCA4(@>(RP1L1(1.27e@17);CRX,FSCN0.02531579 CNGA1(@>(CNGA2,LRRC32,EIF4G3;CNGB1(@>(PABPC3,PDE6B,PDE6B,TXNL4A;IMPDH1(@>(PRPF6,PABPC3;PDE6A(@>(TXNL4A;PDE6G(@>(PRPF6,EIF4G3;PRPF3(@>(PABPC3;PRPF31(@>(TXNL4A;PRPF8(@>(PRPF6,U2AF1,U2AF1;RP9(@>(PABPC3;SNRNP200(@>(TXNL4A1 Haemophilia LUAD 0.7 F8 0.071266219 VWF(@>(Platelet(Aggregation(Inhibitor(Pathway,(Pharmacodynamics(COL3A1,COL4A2,FGB,PLCB1);F8,F9(@>(intrinsic(prothrombin(activation(pathway(F8,PROS1,COL4A2,FGB)1 0.7955 0.52680952 Chronic(Granulomatous(Disease LUSC 1 1 0.067971546 NCF2,NCF4,CYBB,CYBA(@>(B2M(1.31e@02);NCF2,NCF4,CYBB,CYBA(@>(MARCH1(8.59e@03);NCF2,NCF4,CYBB(@>(USP25(1.12e@02);NCF2,NCF4,CYBB,CYBA(@>(NFE2L2(8.87e@03);NCF2,NCF4,CYBB,CYBA(@>(LYZ(2.21e@02);NCF2,NCF4,CYBB,CYBA(@>(PTEN(1.06e@02);NCF2,NCF4,CYBB,CYBA(@>(HDAC10(1.05e@02);NCF2,NCF4,CYBB,CYBA(@>(KDM5A(1.96e@02);NCF2,NCF4,CYBB,CYBA(@>(NFATC1(1.20e@02);NCF4,CYBB,CYBA(@>(CHKB(2.32e@02);NCF2,NCF4,CYBB,CYBA(@>(TRABD(1.01e@02);NCF2,NCF4,CYBB,CYBA(@>(CREBBP(1.06e@02);NCF2,NCF4,CYBB,CYBA(@>(ODF3B(1.41e@02);NCF2,NCF4,CYBB,CYBA(@>(CTDP1(9.25e@03);NCF2,NCF4,CYBB,CYBA(@>(REL(9.24e@03);NCF2,NCF4,CYBB,CYBA(@>(KDM6A(1.20e@02);NCF2,NCF4,CYBB,CYBA(@>(METRNL(9.24e@03);NCF2,NCF4,CYBB,CYBA(@>(PQLC1(8.59e@03);NCF2,NCF4,CYBB,CYBA(@>(LPAR6(1.21e@02);NCF2,NCF4,CYBB,CYBA(@>(BID(1.21e@02);NCF2,NCF4,CYBB,CYBA(@>(RB1(1.92e@02);NCF2,NCF4,CYBB,CYBA(@>(PIM3(8.92e@03);NCF4(@>(NINJ2(1.01e@02);NCF2,NCF4,CYBB,CYBA(@>(EVI2A(1.71e@02);NCF2,NCF4,CYBB,CYBA(@>(EVI2B(9.25e@03);NCF2,NCF4(@>(EXOC3(2.80e@02);NCF2,NCF4,CYBB,CYBA(@>(NOTCH1(8.21e@03);NCF2,NCF4,CY1 1 Congenital(Ichthyosis LUSC 1 1 0.006200982 ALOX12B,ALOXE3,SPINK5,CSTA,KRT2,ABCA12,TGM1(@>(CERS3(3.87e@04);CSTA,NIPAL4,LIPN,ABHD5(@>(NFE2L2(3.14e@02);ALOX12B,SPINK5,CSTA,TGM1(@>(PAX9(3.14e@02);ALOX12B,ALOXE3,CSTA,TGM1(@>(PARD6G(4.70e@02);CSTA,NIPAL4,LIPN,ABHD5(@>(NOTCH1(2.92e@02);ABCA12,TGM1(@>(EGFR(3.98e@02)0.79744737 1 Diamond@Blackfan(Anemia LUSC 1 0.959794083 0.002784462 RPS26(@>(CDKN2A(9.26e@03);(@>(YEATS4(1.30e@03);RPS26,RPS19,RPS7(@>(TP53(1.51e@03);RPS26,RPS19,RPL35A,RPS7(@>(PDCD6(1.33e@04);RPS26,RPS19,RPS10,RPL35A,RPS7(@>(TXNL4A(3.08e@02);(@>(NMU(2.86e@02);RPL5,RPS7(@>(TTF2(4.93e@02);RPS26,RPS19,RPS7(@>(TYMS(2.02e@03);RPS26,RPS7(@>(TRIP13(9.59e@03)1 1 Spinocerebellar(Ataxia LUSC 1 0.977629639 0.01980021 JPH3,CACNA1A,POLG,ATXN2,SYNE1,SPTBN2,TTBK2,KCNC3,PRKCG,FGF14,PPP2R2B(@>(SBF1(9.84e@03);JPH3,ZNF592,ATM,TTBK2,KCNC3,ITPR1,SETX,PPP2R2B(@>(MARCH1(3.18e@02);JPH3,CACNA1A,SPTBN2,TTBK2,KCNC3,ATXN10,PRKCG,FGF14,SYT14,PPP2R2B(@>(L1CAM(1.18e@02);JPH3,ZNF592,CACNA1A,ATXN2,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B,ATXN1(@>(CCSER1(2.45e@03);JPH3,ZNF592,CACNA1A,ATXN2,PDYN,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,AFG3L2,SYT14,PPP2R2B(@>(PARK2(2.80e@03);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(MAPK8IP2(2.90e@03);TDP1,ATM,TBP,NOP56,SYT14,C10orf2(@>(CCDC77(1.54e@02);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(CLCN4(2.99e@03);POLG,ATXN2,TBP,NOP56,AFG3L2,C10orf2(@>(BRD9(2.95e@03);ZNF592,POLG,ATXN7,ATXN2,TDP1,SYNE1,ATM,TTBK2,TBP,ITPR1,SETX,ATXN1(@>(BRD1(3.18e@02);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(CSMD3(2.99e@03);JPH3,CACNA1A1 0.73607576 Combined(Heart(and(Skeletal(Defects LUSC 0.6 CREBBP 3.46E@14 CREBBP(@>(nfat(and(hypertrophy(of(the(heart((NFATC1,CREBBP,PIK3CA);CREBBP(@>(inhibition(of(huntingtons(disease(neurodegeneration(by(histone(deacetylase(inhibitors(CREBBP);EP300,CREBBP(@>(Direct(p53(effectors(TP53,BID,EGFR,RB1,PTEN,CREBBP);EP300,CREBBP(@>(p53(pathway(CDKN2A,TP53,CREBBP);CREBBP(@>(Notch@HLH(transcription(pathway(CREBBP);EP300,CREBBP(@>(mechanism(of(gene(regulation(by(peroxisome(proliferators(via(ppara(FAT1,RB1,CREBBP);CREBBP(@>(regulation(of(transcriptional(activity(by(pml(TP53,CREBBP,RB1);EP300,CREBBP(@>(E2F(transcription(factor(network(TYMS,CDKN2A,CREBBP,RB1);EP300,CREBBP(@>(il@7(signal(transduction(PIK3CA,CREBBP);EP300(@>(p73(transcription(factor(network(MAPK11,CDK6,RB1,WWOX);EP300,CREBBP(@>(Glucocorticoid(receptor(regulatory(network(NFATC1,TP53,MAPK11,CREBBP);CREBBP(@>(Signaling(events(mediated(by(Stem(cell(factor(receptor((c@Kit)(PIK3CA,CREBBP,PTEN);EP300(@>(ATF@2(transcription(factor(network(MAPK11,RB1,NF1);CREBBP(@>(Signaling(events(mediated(by(TCPTP(PIK3CA,EGFR,CREBBP);EP300,CREBBP(@>(F1 0.0925 CREBBP(@>(TP53,TP53;EP300(@>(CREBBP1 Neurofibromatosis LUSC 0.53 NF1 0.236599603 0.719411434 0 NF1(@>(EVI2A,CDKN2A;NF2(@>(NF11 Severe(Combined(Immunodeficiency LUSC 1 0.23074305 0.001138137 ZAP70,IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(B2M(6.47e@03);IL2RG,CIITA,JAK3,ADA,PNP,RFX5,PTPRC,DCLRE1C(@>(PLEKHO1(2.51e@02);ZAP70,IL2RG,CIITA,JAK3,PTPRC,IL7R,CD3D,DCLRE1C(@>(BRD1(1.02e@02);(@>(YEATS4(1.46e@02);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(HDAC10(2.23e@04);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(KDM5A(8.24e@03);IL2RG,JAK3,ADA,PNP,RFX5,PTPRC(@>(NFATC1(1.79e@02);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(FOXP1(2.39e@02);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(CHKB(1.16e@03);ZAP70,IL2RG,CIITA,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(TRABD(1.29e@04);RFXAP,RFX5,DCLRE1C(@>(PRDM15(8.47e@03);ZAP70,IL2RG,JAK3,PNP,RFX5,PTPRC,IL7R,DCLRE1C(@>(CREBBP(3.99e@02);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(CTDP1(3.26e@03);IL2RG,CIITA,JAK3,ADA,PNP,RFX5,PTPRC(@>(REL(3.66e@02);ZAP70,IL2RG,CIITA,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(KDM6A(2.09e@02);ZAP70,IL2RG,RFX5,DCLRE1C(@>(ZBED4(2.61e@05);ADA,AK2,DCLRE1C(@>(CDK6(2.85e@04);ZA1 1 Specified(Hamartoses LUSC 0.63 PTEN 8.39E@06 PTEN(@>(skeletal(muscle(hypertrophy(is(regulated(via(akt@mtor(pathway(PIK3CA,PTEN);PTEN(@>(regulation(of(eif@4e(and(p70s6(kinase(PIK3CA,PTEN);PTEN(@>(mtor(signaling(pathway(PTEN,PIK3CA);PTEN(@>(Direct(p53(effectors(TP53,BID,EGFR,RB1,PTEN,CREBBP);VHL(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53);PTEN(@>(RhoA(signaling(pathway(PTEN,MAPK12,SLC9A3);PTEN(@>(pten(dependent(cell(cycle(arrest(and(apoptosis(PTEN,PIK3CA);PTEN(@>(Signaling(events(mediated(by(Stem(cell(factor(receptor((c@Kit)(PIK3CA,CREBBP,PTEN)0.619582187 0 PTEN(@>(TTF2;SDHB(@>(EGFR;SDHD(@>(CDKN2A;STK11(@>(TP531 Li(Fraumeni(and(Related(Syndromes LUSC 0.03 CDKN2A,TP53 5.79E@15 TP53(@>(Fluoropyrimidine(Pathway,(Pharmacodynamics(TYMS,TP53);TP53(@>(chaperones(modulate(interferon(signaling(pathway(TP53,RB1);TP53(@>(apoptotic(signaling(in(response(to(dna(damage(BID,TP53);TP53(@>(Direct(p53(effectors(TP53,BID,EGFR,RB1,PTEN,CREBBP);TP53,CDKN2A,CHEK2(@>(p53(pathway(CDKN2A,TP53,CREBBP);CDKN2A(@>(C@MYC(pathway(CDKN2A,FBXW7);TP53(@>(estrogen(responsive(protein(efp(controls(cell(cycle(and(breast(tumors(growth(TP53,CDK6);CDKN2A,TP53(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53);TP53(@>(telomeres(telomerase(cellular(aging(and(immortality(TP53,RB1);TP53(@>(btg(family(proteins(and(cell(cycle(regulation(TP53,RB1);TP53(@>(Transcriptional((activation(of((cell(cycle(inhibitor(p21(TP53);TP53(@>(p53(signaling(pathway(TP53,RB1);TP53(@>(regulation(of(transcriptional(activity(by(pml(TP53,CREBBP,RB1);TP53(@>(rb(tumor(suppressor/checkpoint(signaling(in(response(to(dna(damage(TP53,RB1);CDKN2A(@>(E2F(transcription(factor(network(TYMS,CDKN2A,CREBBP,RB1);TP53(@>(Glucocorticoid(receptor0.651962919 0.02405 CDKN2A(@>(PTEN;CHEK2(@>(CDK6,TP53,PTEN;TP53(@>(CREBBP1 Lipoprotein(Deficiencies LUSC 1 0.885354678 0.097427692 APOB,LCAT,APOA1(@>(ENOSF1(1.81e@02);MTTP,APOB,LCAT,SAR1B,APOA1(@>(PROS1(1.81e@02);(@>(AKR1C2(2.16e@02);APOB,LCAT,APOA1(@>(SLC6A12(2.16e@02);MTTP,APOB,LCAT,APOA1(@>(SELO(1.87e@02)1 1 Retinitis(Pigmentosa LUSC 1 EYS 1 5.69E@14 KLHL7,SPATA7,CRB1,MERTK,PDE6B,CERKL,PRCD,FAM161A(@>(CLCN4(4.22e@02);TTC8,CRX,FSCN2,RDH12,PRPH2,RHO,CNGB1,CRB1,NRL,PDE6B,PDE6G,CERKL,GUCA1B,TULP1,NR2E3,ROM1,USH2A,RP1,PRCD,IMPG2,C2orf71,SAG,RBP3,FAM161A(@>(EYS(3.56e@16);TTC8(@>(COLEC12(1.19e@03);TTC8,CRX,FSCN2,PRPH2,CNGB1,KLHL7,SPATA7,CRB1,MERTK,PDE6B,CERKL,TULP1,ROM1,PRCD,IMPG2,C2orf71,RBP3,FAM161A(@>(KCNIP4(1.35e@03);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,CRB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,PROM1,CNGA1,RP1,PRCD,C2orf71,RPE65,SAG,RBP3,FAM161A,ABCA4(@>(CLUL1(2.31e@14);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,CRB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,CNGA1,RP1,PRCD,C2orf71,RPE65,SAG,RBP3,FAM161A,ABCA4(@>(SLC6A13(1.01e@12);TTC8,CRX,LRAT,FSCN2,RDH12,PRPH2,CNGB1,SPATA7,CRB1,MERTK,CERKL,TULP1,ROM1,PROM1,CNGA1,PRCD,IMPG2,C2orf71,RBP3,FAM161A(@>(UNC13B(4.65e@06);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,RP1,PRCD,RPE65,SAG,RBP3,ABCA4(@>(KCNJ13(4.97e@12)1 1 Haemophilia LUSC 1 0.00999936 F8,F9(@>(intrinsic(prothrombin(activation(pathway(COL4A5,PROS1)1 1 1 Chronic(Granulomatous(Disease PRAD 1 1 0.078898011 NCF2,NCF4,CYBA(@>(GPS2(1.49e@02);NCF2,NCF4,CYBB,CYBA(@>(SPOPL(1.38e@02);NCF2,NCF4,CYBB,CYBA(@>(COTL1(3.89e@02);(@>(CRISPLD2(4.13e@02);NCF2,NCF4,CYBB,CYBA(@>(TMUB2(1.62e@02);NCF2(@>(MFI2(2.93e@02);CYBA(@>(ARRDC1(2.54e@02);NCF2,NCF4,CYBB,CYBA(@>(HELZ2(1.23e@02);NCF2,NCF4,CYBB,CYBA(@>(PTEN(1.32e@02);NCF2,NCF4,CYBB(@>(PAK2(1.35e@02);NCF2,NCF4,CYBB,CYBA(@>(NFATC1(1.34e@02);NCF2,NCF4,CYBB,CYBA(@>(CRTC2(1.42e@02);NCF2,NCF4,CYBB,CYBA(@>(ADRBK1(1.57e@02);NCF2,NCF4,CYBA(@>(ZGPAT(1.40e@02);NCF2,NCF4(@>(EGR3(2.34e@02);NCF2,NCF4,CYBA(@>(GPR160(1.26e@02);NCF2,CYBB,CYBA(@>(HNMT(1.21e@02);NCF2,NCF4,CYBB(@>(SENP5(3.88e@02);NCF2,NCF4,CYBB,CYBA(@>(DNAJC5(1.35e@02);NCF2,NCF4,CYBB,CYBA(@>(ANKRD13D(1.27e@02);NCF2,NCF4,CYBB,CYBA(@>(CTDP1(1.30e@02);NCF4,CYBA(@>(CDKN1B(3.93e@02);CYBA(@>(POLD4(2.93e@02);CYBA(@>(RPS27(2.87e@02);NCF2,NCF4,CYBB,CYBA(@>(TPD52L2(1.12e@02);NCF2,NCF4,CYBA(@>(GMEB2(1.40e@02);NCF2,NCF4,CYBB,CYBA(@>(PQLC1(1.16e@02);NCF2,NCF4,CYBB,CYBA(@>(RGS19(2.46e@02);NCF2,NCF4,CYBB,CYBA(@>(TOR4A(1.38e@02);NCF2,NCF4,CYBB(@>(R1 1 Glycogenosis PRAD 1 1 0.078728665 PGAM2,PHKB,PHKA2,AGL(@>(SLC25A30(1.28e@02)1 1 Congenital(Ichthyosis PRAD 1 1 0.02654965 ALOX12B,SPINK5,CSTA,TGM1(@>(C9orf169(2.49e@03);ALOX12B,ALOXE3,CSTA,TGM1(@>(PARD6G(3.82e@02);ALOXE3,CSTA,ABCA12,TGM1(@>(ARRDC1(2.44e@02);NIPAL4,KRT2,LIPN,ABCA12(@>(NRARP(4.60e@03);SPINK5,CSTA,NIPAL4,ABCA12,TGM1(@>(PTK6(4.60e@03)1 1 Disorders(of(Phosphorous(Metabolism PRAD 0.69 SLC34A3 0.000406927 SLC34A3(@>(Type(II(Na+/Pi(cotransporters(SLC34A3) 1 1 1 Spinocerebellar(Ataxia PRAD 1 1 0.027197561 APTX,ATXN7,TDP1,SYNE1,ATM,TTBK2,TBP,ITPR1,SETX,SYT14,C10orf2,PPP2R2B(@>(CASP8AP2(3.80e@03);JPH3,ATXN2,PDYN,SYNE1,ATM,SPTBN2,TTBK2,KCNC3,PRNP,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B,ATXN1(@>(CREBL2(1.97e@02);JPH3,APTX,CACNA1A,ATXN2,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,PRKCG,FGF14,SYT14,PPP2R2B(@>(DLG1(5.25e@03);JPH3,TDP1,KCNC3,FGF14(@>(SAMD10(1.32e@02);JPH3,CACNA1A,ATXN2,PDYN,SPTBN2,TTBK2,KCNC3,PRKCG,FGF14,SYT14,PPP2R2B(@>(NSMF(9.90e@03);JPH3,ZNF592,CACNA1A,ATXN7,ATXN2,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,FGF14,SETX,PPP2R2B,ATXN1(@>(PHC3(2.61e@03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(GRIN1(3.68e@03);JPH3,CACNA1A,ATXN2,SYNE1,SPTBN2,TTBK2,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(MYT1(2.84e@03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,ATXN10,PRKCG,FGF14,SYT14,PPP2R2B(@>(ZNF285(1.11e@02);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(STMN3(2.84e@03);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN1 1 Specified(Hamartoses PRAD 0.66 PTEN 0.062379972 STK11(@>(Metformin(Pathway,(Pharmacodynamic(SLC2A4,CRTC2);PTEN(@>(RhoA(signaling(pathway(CDKN1B,PTEN);PTEN(@>(pten(dependent(cell(cycle(arrest(and(apoptosis(CDKN1B,PTEN);PTEN(@>(Negative(regulation(of(the(PI3K/AKT(network(PTEN)0.57662662 1 1 Retinitis(Pigmentosa PRAD 1 1 6.10E@15 CRX,EYS,SPATA7,CRB1,MERTK,PDE6B,CERKL,PRCD,IMPG2,C2orf71,RBP3(@>(CREBL2(4.56e@02);CRX,FSCN2,PRPH2,CNGB1,KLHL7,EYS,SPATA7,CRB1,PDE6B,CERKL,TULP1,PRCD,IMPG2,C2orf71,RBP3,FAM161A(@>(MYT1(7.66e@06);SNRNP200,CRB1,PRPF31,CERKL,IDH3B,PRPF8,FAM161A(@>(UCKL1(1.68e@03);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,EYS,CRB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,PROM1,CNGA1,RP1,PRCD,C2orf71,RPE65,SAG,RBP3,FAM161A,ABCA4(@>(KCNG4(2.10e@17);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,EYS,CRB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,PROM1,CNGA1,RP1,PRCD,C2orf71,RPE65,SAG,RBP3,FAM161A,ABCA4(@>(SAMD7(1.45e@17);RPGR,CA4,PRPF3,BEST1,RP2,TOPORS,SEMA4A(@>(GPR160(1.80e@02);SNRNP200,EYS(@>(THSD7B(1.89e@05);(@>(WFDC1(1.49e@05);(@>(NXPH2(5.64e@03);TTC8,RDH12,SPATA7,MERTK,CNGA1,FAM161A(@>(ZFHX3(1.68e@03);CRX,FSCN2,RDH12,SNRNP200,PRPH2,CNGB1,KLHL7,EYS,SPATA7,CRB1,PDE6B,CERKL,PRPF3,TULP1,ROM1,PRCD,IMPG2,C2orf71,TOPORS,RBP3,FAM161A(@>(PCMTD2(7.66e@06);ZNF513,KLHL7,SPATA7,CRB1,PDE6B,CERKL,FAM161A(@>(TMEM145(2.0.34981818 0.26722222 Long(QT(Syndrome READ 0.69 CACNA1C 3.14E@21 CACNA1C(@>(Nicotine(Pathway((Chromaffin(Cell),(Pharmacodynamics(CACNA1C);CACNA1C(@>(Sympathetic(Nerve(Pathway((Pre@(and(Post@(Ganglionic(Junction)(CACNA1C,TH);CACNA1C(@>(Anti@diabetic(Drug(Potassium(Channel(Inhibitors(Pathway,(Pharmacodynamics(PDX1,CACNA1C,INS)1 1 1 Chronic(Granulomatous(Disease READ 1 1 0.07714767 (@>(CRLF2(3.23e@02);NCF2,NCF4,CYBB,CYBA(@>(PRPF3(2.17e@02);NCF2,NCF4,CYBB,CYBA(@>(CACNA2D4(1.29e@02);NCF2(@>(KRAS(4.79e@02);NCF4,CYBA(@>(TRAPPC2(2.54e@02);NCF2,NCF4,CYBB,CYBA(@>(PLAGL2(1.33e@02);NCF2,NCF4,CYBB,CYBA(@>(CSF2RA(1.45e@02);NCF2,CYBB,CYBA(@>(TCF7L2(1.37e@02);NCF2,NCF4,CYBA(@>(TCEANC(2.36e@02);NCF2,NCF4,CYBB(@>(CTNNBL1(1.42e@02);NCF2,NCF4(@>(RNF40(3.72e@02);NCF2,NCF4,CYBB,CYBA(@>(IRF2(1.53e@02);NCF2,NCF4,CYBB,CYBA(@>(CR1(1.22e@02);NCF2,NCF4,CYBB,CYBA(@>(FBRS(1.23e@02);NCF2,CYBB,CYBA(@>(HS3ST3B1(1.16e@02);NCF4(@>(C17orf103(2.44e@02);NCF2,NCF4,CYBB,CYBA(@>(IL3RA(1.69e@02);(@>(ADK(3.11e@02);NCF2,CYBB(@>(RAB9A(1.57e@02);NCF2,CYBB(@>(EMP1(4.04e@02);NCF2(@>(SRCAP(3.53e@02);NCF2,NCF4,CYBB,CYBA(@>(RAB39A(1.60e@02);NCF2,NCF4,CYBB,CYBA(@>(MAP2K3(2.17e@02);NCF2,NCF4,CYBB,CYBA(@>(MOSPD2(1.16e@02);NCF2,NCF4,CYBB,CYBA(@>(MCL1(1.19e@02);NCF2(@>(P2RY8(1.27e@02)1 1 Glycogenosis READ 0.86 PHKG2 6.77E@12 AGL,PYGM,PHKA1,PHKB,PHKA2,PHKG2(@>(Glycogen(breakdown((glycogenolysis)(PHKG2,GYG2)0.372442654 1 0.36782353 Inherited(Anomalies(of(the(Skin READ 1 0.004631074 TERT(@>(HIF@1@alpha(transcription(factor(network(SMAD4,MCL1,SMAD3);TERT(@>(telomeres(telomerase(cellular(aging(and(immortality(TP53,KRAS);TERT(@>(Validated(targets(of(C@MYC(transcriptional(activation(SMAD4,TP53,SMAD3)0.10458489 1 1 Spinocerebellar(Ataxia READ 1 ATXN10 0.008075087 CACNA1A(@>(Sympathetic(Nerve(Pathway((Pre@(and(Post@(Ganglionic(Junction)(CACNA1C,TH);PRNP(@>(Glypican(1(network(HCK,FGFR1);CACNA1A(@>(Anti@diabetic(Drug(Potassium(Channel(Inhibitors(Pathway,(Pharmacodynamics(PDX1,CACNA1C,INS);TBP(@>(Validated(targets(of(C@MYC(transcriptional(repression(ERBB2,SMAD4,SMAD3)0.013136049 ATXN7,TDP1,SYNE1,ATM,TTBK2,TBP,ITPR1,PPP2R2B(@>(FHIT(3.07e@03);(@>(PPP2R3B(4.28e@02);JPH3,ATXN2,TTBK2,TBP,PPP2R2B(@>(CDRT4(4.76e@02);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B(@>(GPM6B(3.44e@03);ZNF592,POLG,ATXN7,TDP1,ATM,TBP,ITPR1,SETX,ATXN1(@>(PRPF3(8.48e@03);JPH3,ZNF592,CACNA1A,ATXN2,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B,ATXN1(@>(CCSER1(2.41e@03);JPH3,ZNF592,CACNA1A,ATXN7,ATXN2,PDYN,TDP1,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SETX,SYT14,PPP2R2B,ATXN1(@>(WHSC1L1(1.11e@03);JPH3,ZNF592,CACNA1A,ATXN2,PDYN,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,AFG3L2,SYT14,PPP2R2B(@>(PARK2(2.44e@03);JPH3,SPTBN2,TTBK2,KCNC3,ITPR1,PRKCG,FGF14,PPP2R2B(@>(ENSA(4.21e@02);JPH3,ZNF592,CACNA1A,ATXN2,PDYN,SYNE1,ATM,SPTBN2,TTBK2,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B,ATXN1(@>(ZNF785(1.51e@03);JPH3,ZNF592,CACNA1A,ATXN7,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,FGF14,SETX,SYT14,PPP2R2B,ATXN1(@>(CUL5(1.42e@03);POLG,ATXN7,TDP1,SYNE1,ATM,TBP,ITPR1,1 1 Severe(Combined(Immunodeficiency READ 1 1 0.017157155 IL2RG,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,DCLRE1C(@>(PRPF3(1.29e@02);ZAP70,IL2RG,CIITA,JAK3,RFXAP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(OFD1(1.32e@03);RFXAP,DCLRE1C(@>(FNTA(4.56e@02);PNP(@>(CACNA2D4(4.19e@02);ZAP70,IL2RG,JAK3,RFXAP,IL7R,CD3D,DCLRE1C(@>(CSTF2T(1.10e@02);(@>(DDX47(3.75e@02);IL2RG,JAK3,RFXANK,RFX5,PTPRC(@>(PRR14(6.34e@03);IL2RG,JAK3,ADA,PNP,RFX5,PTPRC,DCLRE1C(@>(PLAGL2(1.07e@02);ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(TCEANC(2.91e@03);ZAP70,IL2RG,JAK3,RFXAP,RFX5,IL7R,CD3D,DCLRE1C(@>(ASXL1(1.37e@03);PNP(@>(CTNNBL1(3.18e@02);IL2RG,JAK3,ADA,PNP,PTPRC(@>(HCK(4.98e@02);ZAP70,IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,IL7R,DCLRE1C(@>(IRF2(9.30e@03);IL2RG,JAK3,ADA,PNP,PTPRC(@>(HS3ST3B1(1.89e@02);JAK3,PTPRC(@>(C17orf103(4.06e@02);RFX5(@>(GPRC5D(1.63e@02);CIITA(@>(FLT3(1.41e@03);RFXAP,AK2,DCLRE1C(@>(FANCB(2.62e@02);CIITA,RFX5,DCLRE1C(@>(ADK(1.46e@02);PNP,PTPRC(@>(P2RY8(2.19e@03);RFXAP(@>(ANK1(4.68e@02)1 1 Lipoprotein(Deficiencies READ 1 1 0.075339054 MTTP,APOB,LCAT,SAR1B,APOA1(@>(KLC4(3.60e@02);APOB,LCAT,APOA1(@>(FCN3(2.07e@02);MTTP(@>(INS@IGF2(2.31e@02);APOB,ABCA1(@>(HS3ST3B1(2.70e@02);(@>(ADK(3.79e@02);MTTP,APOB,LCAT,APOA1(@>(ARSD(1.18e@02);MTTP,APOB,LCAT,SAR1B,APOA1(@>(ARSE(2.07e@02)1 1 Disorders(of(Urea(Cycle(Metabolism READ 1 1 0.088390959 ASS1,NAGS,ARG1,ASL,CPS1(@>(KLC4(4.33e@02);NAGS,ARG1,ASL,CPS1(@>(FCN3(2.80e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(ARSD(1.57e@02);ASS1,NAGS,ARG1,ASL,CPS1(@>(ARSE(2.80e@02)1 1 Retinitis(Pigmentosa READ 1 PRPF3 1 4.04E@13 (@>(IMMP2L(2.15e@05);(@>(DHRSX(1.34e@03);EYS,CERKL(@>(NKX6@3(1.42e@02);RPGR,CRX,FSCN2,RDH12,PRPH2,CNGB1,EYS,CRB1,TULP1,ROM1,RP2,PRCD,IMPG2,C2orf71,RBP3,SEMA4A(@>(CACNA2D4(1.12e@08);FSCN2,PRPH2,RHO,CNGB1,PDE6A,PDE6G,GUCA1B,RLBP1,RGR,NR2E3,ROM1,CNGA1,RP1,SAG,ABCA4(@>(EGFL6(4.38e@09);CRX,FSCN2,RDH12,PRPH2,CNGB1,EYS,CRB1,CERKL,TULP1,ROM1,PRCD,IMPG2,C2orf71,RBP3(@>(ASMT(1.12e@08);TTC8,CRX,FSCN2,RDH12,PRPH2,CNGB1,KLHL7,EYS,SPATA7,CRB1,MERTK,PDE6B,CERKL,TULP1,ROM1,PRCD,IMPG2,C2orf71,RBP3,FAM161A(@>(KIAA1467(6.02e@04);(@>(ZBED1(3.70e@03);TTC8,CRX,FSCN2,PRPH2,CNGB1,KLHL7,EYS,SPATA7,CRB1,TULP1,PRCD,IMPG2,C2orf71,RBP3,FAM161A(@>(GSG1(6.99e@07);CRX,FSCN2,RDH12,PRPH2,RHO,CNGB1,NRL,PDE6B,PDE6G,GUCA1B,TULP1,NR2E3,ROM1,PROM1,CNGA1,RP1,PRCD,C2orf71,SAG,RBP3,FAM161A,ABCA4(@>(KERA(5.05e@15);FAM161A(@>(SHOX(9.94e@11)1 0.40083333 Dopa@Responsive(Dystonia READ 0.62 TH 2.34E@32 TH(@>(Alpha@synuclein(signaling(HCK,TH,PARK2);TH(@>(Sympathetic(Nerve(Pathway((Pre@(and(Post@(Ganglionic(Junction)(CACNA1C,TH)1 1 1 Congenital(Ectodermal(Dysplasia SKCM 1 0.28548436 0.000106457 (@>(FAM58A(9.62e@03);LAMB3,ITGB4,PLEC,KRT5,LAMC2(@>(ANAPC15(2.97e@03);(@>(EIF3D(1.58e@03);ITGA6,LAMB3,ITGB4,GJB6,COL17A1,PLEC,COL7A1,KRT5,KRT14,LAMC2,LAMA3(@>(PTK6(2.00e@06);LAMB3,ITGB4,KRT5(@>(MYC(2.50e@04);(@>(CDKN2A(2.84e@02);PLEC(@>(PHGDH(3.27e@02);PLEC(@>(RPL13(4.57e@03);(@>(CCND1(2.48e@02);LAMB3,ITGB4,PLEC,KRT5(@>(PPDPF(2.47e@03);(@>(LSM12(2.57e@02);(@>(SLC25A39(1.42e@02);(@>(MRP63(4.21e@02);(@>(SRMS(2.88e@02);ITGA6,LAMB3,ITGB4,COL17A1,KRT5,KRT14,LAMC2,LAMA3(@>(TDRP(5.01e@04);PLEC(@>(GRN(1.79e@02);(@>(TP53(1.56e@02);(@>(TSPAN31(1.19e@02);(@>(HDAC3(3.26e@03);ITGB4,PLEC,KRT5(@>(ACD(5.04e@03);(@>(KRT78(1.13e@02);GJB6(@>(TCHHL1(9.83e@03);PLEC(@>(DEF8(2.72e@02);ITGA6,COL17A1,COL7A1,KRT14(@>(DSG1(1.08e@05);LAMB3,ITGB4,PLEC,KRT5,KRT14,LAMC2(@>(TNFRSF6B(6.12e@04);(@>(RPTN(3.38e@04);PLEC(@>(TPD52L2(4.98e@02);LAMB3,ITGB4,PLEC,COL7A1,KRT5(@>(CHMP1A(1.44e@02);(@>(DYNAP(9.82e@03);ITGB4,PLEC,COL7A1(@>(SLC2A4RG(6.70e@03);(@>(FOLR3(2.43e@03)1 1 Chronic(Granulomatous(Disease SKCM 1 1 0.078763609 NCF2,NCF4,CYBB,CYBA(@>(ZFX(1.54e@02);NCF2,NCF4,CYBB,CYBA(@>(HELZ2(1.79e@02);NCF2,NCF4(@>(KIAA1257(2.33e@02);NCF2(@>(STK19(3.34e@02);NCF2,NCF4,CYBB,CYBA(@>(B2M(2.08e@02);NCF2,NCF4,CYBB(@>(VPS9D1(1.84e@02);NCF2,NCF4,CYBB,CYBA(@>(DDX3X(1.31e@02);NCF2,NCF4,CYBB,CYBA(@>(TMUB2(1.76e@02);NCF2,NCF4,CYBB,CYBA(@>(CLEC2B(2.70e@02);NCF2,NCF4,CYBB,CYBA(@>(ADAM8(1.31e@02);NCF2,NCF4,CYBB,CYBA(@>(PTEN(1.49e@02);NCF2,NCF4,CYBB,CYBA(@>(ZNF276(1.43e@02);NCF4(@>(SLC25A39(1.56e@02);NCF2,NCF4,CYBB,CYBA(@>(GNAI2(1.51e@02);NCF2,NCF4,CYBB,CYBA(@>(PPP6C(1.49e@02);NCF2,NCF4,CYBB,CYBA(@>(RPGRIP1(1.45e@02);NCF4,CYBB(@>(POM121(3.52e@02);NCF2,NCF4,CYBA(@>(ZGPAT(1.56e@02);(@>(SERPINB10(1.35e@02);NCF2,NCF4,CYBB,CYBA(@>(GRN(3.32e@02);NCF2,NCF4,CYBB,CYBA(@>(DNAJC5(1.52e@02);NCF2,NCF4,CYBB,CYBA(@>(RBM22(1.44e@02);NCF4(@>(ITGA2B(3.33e@02);CYBB,CYBA(@>(OXA1L(1.33e@02);NCF2,NCF4,CYBB(@>(MPP7(1.54e@02);(@>(SLC4A1(3.39e@02);NCF2,NCF4,CYBB,CYBA(@>(CLEC5A(1.39e@02);NCF2,NCF4,CYBB,CYBA(@>(GPR141(1.49e@02);NCF4,CYBB,CYBA(@>(ITGA4(1.49e@02);NCF2,NCF4,CYB1 1 Congenital(Ichthyosis SKCM 1 1 0.005136028 SPINK5,CSTA,NIPAL4,ABCA12,TGM1(@>(PTK6(3.02e@03);SPINK5,CSTA,NIPAL4,KRT2,LIPN,ABHD5(@>(MPP7(1.32e@02);ALOX12B,SPINK5,CSTA,TGM1(@>(KRT78(5.11e@03);ALOX12B,ABCA12(@>(TCHHL1(3.36e@04);ALOX12B,SPINK5,CSTA,ABCA12,TGM1(@>(DSG1(2.89e@04);ALOX12B,SPINK5,CSTA,ABCA12,TGM1(@>(RPTN(2.89e@04);ALOX12B,SPINK5,CSTA,TGM1(@>(DYNAP(3.02e@03)1 1 Polycystic(Kidney,(Autosomal(Dominant SKCM 1 0.013304666 TSC2(@>(LKB1(signaling(events(RPTOR,MYC,TP53);TSC2(@>(Validated(targets(of(C@MYC(transcriptional(repression(CCND1,HDAC3,EP300,MYC)0.736679678 1 1 Inherited(Anomalies(of(the(Skin SKCM 1 TERT 5.99E@19 TERT(@>(erk1/erk2(mapk(signaling(pathway(MYC,TERT);TERT(@>(IL2(signaling(events(mediated(by(PI3K(MYC,TERT,RAC1);TERT(@>(overview(of(telomerase(protein(component(gene(htert(transcriptional(regulation(MYC,TERT,TP53);TERT(@>(telomeres(telomerase(cellular(aging(and(immortality(MYC,TERT,TP53);KRT1,TERT(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(TERT,CCND1,CDKN2A,MITF,MYC,EP300);TERT(@>(role(of(nicotinic(acetylcholine(receptors(in(the(regulation(of(apoptosis(TERT,FASLG);TERT(@>(Validated(targets(of(C@MYC(transcriptional(activation(TERT,TP53,UBTF,CDK4,MYC,EP300);TINF2,TERT,DKC1(@>(Regulation(of(Telomerase(CCND1,MYC,TERT,SAP18,ACD)0.164716031 1 1 Combined(Heart(and(Skeletal(Defects SKCM 0.64 EP300 2.01E@19 EP300(@>(cell(cycle:(g2/m(checkpoint(EP300,TP53,MYT1);EP300(@>(Validated(transcriptional(targets(of(AP1(family(members(Fra1(and(Fra2(CCND1,EP300,CDKN2A);EP300,CREBBP(@>(acetylation(and(deacetylation(of(rela(in(nucleus(HDAC3,EP300);EP300(@>(Notch(signaling(pathway(CCND1,MYC,EP300,NOTCH2,DNER);EP300(@>(Validated(nuclear(estrogen(receptor(alpha(network(CCND1,KLRC3,EP300,MYC);EP300(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(TERT,CCND1,CDKN2A,MITF,MYC,EP300);EP300(@>(Validated(targets(of(C@MYC(transcriptional(repression(CCND1,HDAC3,EP300,MYC);EP300,CREBBP(@>(E2F(transcription(factor(network(CCNE2,EP300,MYC,CDKN2A);EP300,CREBBP(@>(il@7(signal(transduction(ITGA2B,EP300);EP300,CREBBP(@>(Validated(targets(of(C@MYC(transcriptional(activation(TERT,TP53,UBTF,CDK4,MYC,EP300);EP300(@>(hypoxia(and(p53(in(the(cardiovascular(system(EP300,TP53);EP300(@>(ATF@2(transcription(factor(network(CCND1,EP300,CDK4,NF1);EP300(@>(melanocyte(development(and(pigmentation(pathway(MITF,EP300);EP300,CREBBP(@>1 0.32066667 1 Specified(Anomalies(of(the(Musculoskeletal(SystemSKCM 1 MITF 0.097075171 MITF,SNAI2(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(TERT,CCND1,CDKN2A,MITF,MYC,EP300);MITF(@>(IL6@mediated(signaling(events(MITF,MYC,RAC1);MITF(@>(melanocyte(development(and(pigmentation(pathway(MITF,EP300)1 1 1 Neurofibromatosis SKCM 0.61 NF1 4.53E@05 NF1(@>(Regulation(of(Ras(family(activation(NRAS,RASA2,NF1);NF1(@>(ATF@2(transcription(factor(network(CCND1,EP300,CDK4,NF1)0.719411434 0.222 NF2(@>(CDKN2A 1 Tuberous(Sclerosis SKCM 1 0.013304666 TSC2,TSC1(@>(LKB1(signaling(events(RPTOR,MYC,TP53);TSC2(@>(Validated(targets(of(C@MYC(transcriptional(repression(CCND1,HDAC3,EP300,MYC)0.594661126 1 1 Severe(Combined(Immunodeficiency SKCM 1 1 0.001386824 IL2RG,JAK3,ADA,PNP,PTPRC(@>(HELZ2(4.66e@02);ZAP70,IL2RG,JAK3,PTPRC,IL7R,CD3D,DCLRE1C,RAG2,RAG1(@>(THEMIS(1.76e@04);ZAP70,IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(B2M(6.45e@03);NHEJ1,RFX5,AK2(@>(EIF3D(4.81e@05);ZAP70,IL2RG,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(KCNAB2(3.23e@03);ZAP70,RFXANK,IL7R,CD3D(@>(RPL13(2.74e@03);AK2(@>(TERT(1.68e@02);ZAP70,IL2RG,JAK3,PTPRC,IL7R,CD3D,DCLRE1C(@>(TC2N(5.55e@03);ZAP70,IL2RG,JAK3,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(CLEC2B(3.84e@03);ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(CLEC2D(1.81e@04);IL2RG,JAK3,PNP,RFX5,PTPRC(@>(ADAM8(3.98e@02);ZAP70,IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(ZNF276(3.22e@03);ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(PCED1B(1.76e@03);JAK3(@>(SLC25A39(7.97e@04);IL2RG,CIITA,PNP,RFX5,PTPRC,DCLRE1C(@>(LY86(1.75e@02);ZAP70,IL2RG,JAK3,PNP,RFX5,PTPRC,DCLRE1C(@>(PPP6C(5.65e@03);ZAP70,IL2RG,JAK3,RFXAP,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C(@>(POM121(8.56e@04);ZAP70,IL2RG,JAK3,ADA,PTPRC,IL7R,CD3D(@>(ZGPAT(2.74e@03);PNP,AK2(@>(1 1 Specified(Hamartoses SKCM 0.72 PTEN 0.061006438 STK11(@>(LKB1(signaling(events(RPTOR,MYC,TP53);VHL(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53);STK11(@>(Regulation(of(AMPK(activity(via(LKB1(RPTOR);PTEN(@>(pten(dependent(cell(cycle(arrest(and(apoptosis(PTEN,FASLG)0.561087039 0.13528125 PTEN(@>(TP53;SDHB(@>(CDKN2A;STK11(@>(NRAS,OGDHL,TP530.2405 PTEN(@>(TCEB3C;SDHB(@>(TP53;STK11(@>(CCNE2;VHL(@>(CHGB Li(Fraumeni(and(Related(Syndromes SKCM 0.05 CDKN2A,TP53 2.63E@31 TP53(@>(LKB1(signaling(events(RPTOR,MYC,TP53);TP53,CHEK2(@>(cell(cycle:(g2/m(checkpoint(EP300,TP53,MYT1);CDKN2A(@>(Validated(transcriptional(targets(of(AP1(family(members(Fra1(and(Fra2(CCND1,EP300,CDKN2A);TP53(@>(estrogen(responsive(protein(efp(controls(cell(cycle(and(breast(tumors(growth(CDK4,TP53);TP53(@>(overview(of(telomerase(protein(component(gene(htert(transcriptional(regulation(MYC,TERT,TP53);TP53,CHEK2(@>(role(of(brca1(brca2(and(atr(in(cancer(susceptibility(TP53,FANCA);CDKN2A,TP53(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53);TP53(@>(telomeres(telomerase(cellular(aging(and(immortality(MYC,TERT,TP53);CDKN2A(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(TERT,CCND1,CDKN2A,MITF,MYC,EP300);TP53(@>(btg(family(proteins(and(cell(cycle(regulation(CCND1,TP53);TP53(@>(Transcriptional((activation(of((cell(cycle(inhibitor(p21(TP53);TP53(@>(p53(signaling(pathway(CCND1,CDK4,TP53);TP53(@>(rb(tumor(suppressor/checkpoint(signaling(in(response(to(dna(damage(CDK40.662892321 0.15561765 CDKN2A(@>(PTEN;CHEK2(@>(CDK4,TP53,PTEN;TP53(@>(EP3001 Lipoprotein(Deficiencies SKCM 1 1 0.058815075 APOB,LCAT,APOA1(@>(SPTLC3(3.68e@02);APOB,SAR1B(@>(IDH1(2.34e@02);MTTP,APOB,LCAT,SAR1B,APOA1(@>(UGT2B15(6.62e@03);MTTP,APOB,LCAT,SAR1B,APOA1,ABCA1(@>(C2(1.13e@02);APOB,APOA1(@>(TIMD4(3.68e@02);APOB,APOA1(@>(STAB2(1.22e@02);MTTP,APOB,LCAT,SAR1B,APOA1(@>(APCS(1.13e@02)1 1 Retinitis(Pigmentosa SKCM 1 EYS,CRB1 1 2.00E@13 CERKL,IDH3B(@>(ASB16(5.43e@03);CRX,FSCN2,KLHL7,SPATA7,MERTK,PDE6B,CERKL,ROM1,PRCD,C2orf71,RBP3,FAM161A(@>(SV2B(3.21e@03);CRX,FSCN2,PRPH2,CNGB1,KLHL7,SPATA7,PDE6B,CERKL,TULP1,PRCD,IMPG2,C2orf71,RBP3,FAM161A(@>(MYT1(4.84e@05);TTC8,CRX,FSCN2,RDH12,PRPH2,RHO,CNGB1,KLHL7,SPATA7,NRL,MERTK,PDE6B,PDE6G,CERKL,GUCA1B,TULP1,NR2E3,ROM1,PRCD,IMPG2,C2orf71,SAG,RBP3,FAM161A,ABCA4(@>(CRB1(2.61e@05);CRX,FSCN2,RDH12,PRPH2,CNGB1,CERKL,TULP1,ROM1,PRCD,IMPG2,C2orf71,RBP3(@>(LRTM1(5.49e@09);TTC8,CRX,FSCN2,RDH12,PRPH2,CNGB1,SPATA7,PDE6B,TULP1,ROM1,PRCD,RBP3(@>(FMN1(2.31e@04);CRX,FSCN2,RDH12,PRPH2,RHO,CNGB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,PROM1,CNGA1,RP1,PRCD,C2orf71,SAG,RBP3,FAM161A,ABCA4(@>(PROL1(6.23e@14);KLHL7(@>(NUDT4(2.32e@02);KLHL7,IDH3B(@>(SCN5A(4.86e@02);LRAT,SNRNP200,PROM1(@>(UTF1(4.41e@02);ZNF513,KLHL7,SPATA7,MERTK,PDE6B,PRCD,FAM161A(@>(CHRNA4(1.59e@02);CRX,SNRNP200,CA4,CERKL,TULP1,PROM1,FAM161A(@>(KCNB2(9.21e@04);TTC8,RPGR,SPATA7,CERKL,FAM161A(@>(CHGB(4.86e@02);(@>(MITF(1.67e@04);(@>(SPTBN5(4.63e@0.21645 IDH3B(@>(IDH1,EIF2B1;IMPDH1(@>(SKIV2L;PROM1(@>(DDX3X;PRPF3(@>(RPL13,CRB1,PRPF6,PRPF6,DDX3X;PRPF31(@>(PRPF6,RPGRIP1,MYC;PRPF8(@>(PRPF60.22446667 CA4(@>(SLC4A1,EP300;CRX(@>(RBFOX1;FAM161A(@>(PARK2,RUNDC3A,MYC;IDH3B(@>(TP53,DDX3X,HDAC3,PRPF6,FANCA;IMPDH1(@>(PRPF6,DDX3X;NR2E3(@>(ITGA4;PRPF3(@>(HDAC5;PRPF31(@>(MYC;PRPF8(@>(PRPF6,PHGDH,RPGRIP1,HDAC5;ROM1(@>(ITGA4;RPGR(@>(PARK2;SNRNP200(@>(MYC;TOPORS(@>(PRPF6,UPF3A Hereditary(Hemorrhagic(Telangiectasia SKCM 1 1.86E@06 SMAD4(@>(LKB1(signaling(events(RPTOR,MYC,TP53);SMAD4(@>(Validated(nuclear(estrogen(receptor(alpha(network(CCND1,KLRC3,EP300,MYC);SMAD4(@>(Validated(targets(of(C@MYC(transcriptional(repression(CCND1,HDAC3,EP300,MYC);SMAD4(@>(Validated(targets(of(C@MYC(transcriptional(activation(TERT,TP53,UBTF,CDK4,MYC,EP300)1 1 1 Disorders(of(Aromatic(Amino(Acid(MetabolismSKCM 1 MC1R 1 0.072585714 BLOC1S6,AP3B1,BLOC1S3,HPS5,HPS1(@>(HELZ2(1.61e@02);BLOC1S6,AP3B1,BLOC1S3,HPS5,HPS6,HPS1(@>(TMUB2(3.76e@02);TYR,OCA2,TYRP1,SLC45A2,BLOC1S6,AP3B1,DTNBP1,BLOC1S3,HPS5,HPS1(@>(KCNAB2(9.33e@03);BLOC1S6,AP3B1,BLOC1S3,HPS6,HPS1(@>(GNAI2(2.33e@02);BLOC1S6,BLOC1S3,HPS5,HPS6,HPS1(@>(GRN(1.01e@02);(@>(OXA1L(2.24e@02);BLOC1S6,AP3B1,HPS5(@>(CLEC5A(3.63e@02);BLOC1S6,BLOC1S3,HPS5,HPS6,HPS1(@>(TPD52L2(3.55e@02);TAT,HPD,FAH(@>(C2(3.91e@02);HPD(@>(LIME1(4.55e@02);BLOC1S6,AP3B1,BLOC1S3,HPS5,HPS1(@>(RGS19(3.70e@02);TYR,OCA2,TYRP1,SLC45A2(@>(MITF(2.04e@02);BLOC1S3,HPS6,HPS1(@>(INPPL1(2.36e@02);TYR,OCA2,TYRP1,SLC45A2,BLOC1S3,HPS5,HPS1(@>(MAD1L1(1.01e@02);BLOC1S6,DTNBP1,BLOC1S3,HPS1(@>(TCF25(3.81e@02);BLOC1S6,AP3B1,DTNBP1,BLOC1S3,HPS5,HPS1(@>(SBNO2(2.33e@02);HPS1(@>(FOLR2(1.98e@02)1 1 Chronic(Granulomatous(Disease STAD 1 0.846838069 0.072585714 NCF2,NCF4,CYBB,CYBA(@>(B2M(2.04e@02);NCF2,NCF4,CYBB,CYBA(@>(DIAPH2(1.14e@02);NCF2,NCF4,CYBB,CYBA(@>(TRPS1(9.36e@03);NCF2(@>(KRAS(4.90e@02);CYBA(@>(CLECL1(3.26e@02);NCF2,NCF4,CYBB,CYBA(@>(CLEC2B(2.77e@02);NCF2,NCF4,CYBB,CYBA(@>(PTEN(1.14e@02);NCF2,NCF4,CYBB,CYBA(@>(SNX2(1.75e@02);NCF2,NCF4,CYBB,CYBA(@>(CLEC12A(1.60e@02);NCF4(@>(IL5RA(3.13e@02);NCF2,CYBB,CYBA(@>(RHOA(9.22e@03);NCF2,NCF4,CYBB,CYBA(@>(IRF2(9.69e@03);(@>(KLRF1(3.81e@02);(@>(CLEC1B(3.30e@02);NCF4,CYBA(@>(CD69(3.41e@02);NCF2,NCF4,CYBB,CYBA(@>(DYRK1A(2.15e@02);NCF2,CYBB,CYBA(@>(UAP1L1(1.35e@02);(@>(PLGRKT(3.41e@02);CYBB,CYBA(@>(DPP7(1.97e@02);NCF2,CYBB,CYBA(@>(CD44(9.47e@03);NCF2,NCF4,CYBB,CYBA(@>(CD274(1.06e@02)1 1 Disorders(of(Phosphorous(Metabolism STAD 0.67 SLC34A3 0.003812225 FGF23(@>(Syndecan@2@mediated(signaling(events(RHOA,FGFR2,FGF19);FGF23(@>(Syndecan@3@mediated(signaling(events(FGFR2,FGF19,EGFR);SLC34A3(@>(Type(II(Na+/Pi(cotransporters(SLC34A3);FGF23(@>(FGF(signaling(pathway(CDH1,PIK3CA,FGF19,FGFR2)1 0.11223333 FGF23(@>(FGF19 1 Combined(Heart(and(Skeletal(Defects STAD 1 0.002558855 CREBBP(@>(the(information(processing(pathway(at(the(ifn(beta(enhancer(IRF2,ARID1A);EP300(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(CDH1,APC,AES,CDKN2A);EP300(@>(p73(transcription(factor(network(RNF43,CDK6,WWOX)0.650459398 0.04182609 CREBBP(@>(TP53,TP53,GATA40.26054167 Specified(Hamartoses STAD 0.63 PTEN 1.38E@07 PTEN(@>(skeletal(muscle(hypertrophy(is(regulated(via(akt@mtor(pathway(PIK3CA,PTEN);PTEN(@>(regulation(of(eif@4e(and(p70s6(kinase(PIK3CA,PTEN);PTEN(@>(mtor(signaling(pathway(PTEN,PIK3CA);VHL(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53);PTEN(@>(RhoA(signaling(pathway(PTEN,RHOA,MAP2K4);PTEN(@>(pten(dependent(cell(cycle(arrest(and(apoptosis(PTEN,PIK3CA);PTEN(@>(Negative(regulation(of(the(PI3K/AKT(network(PTEN)0.719559223 0 PTEN(@>(PIK3CA;STK11(@>(EGFR1 Li(Fraumeni(and(Related(Syndromes STAD 0.03 CDKN2A,TP53 1.68E@20 CDKN2A(@>(C@MYC(pathway(CDKN2A,FBXW7);TP53(@>(estrogen(responsive(protein(efp(controls(cell(cycle(and(breast(tumors(growth(TP53,CDK6);CDKN2A,TP53(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(CDKN2A,TP53);TP53(@>(telomeres(telomerase(cellular(aging(and(immortality(TP53,KRAS);CDKN2A(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(CDH1,APC,AES,CDKN2A);TP53(@>(BARD1(signaling(events(CCNE1,TP53);TP53(@>(Transcriptional((activation(of((cell(cycle(inhibitor(p21(TP53);TP53,CHEK2(@>(PLK3(signaling(events(CCNE1,TP53);TP53(@>(p53(signaling(pathway(CCNE1,TP53);TP53(@>(p75(NTR)@mediated(signaling(PIK3CA,RHOA,TP53)0.693511719 0 CDKN2A(@>(PTEN;CHEK2(@>(CDK6,TP53,SMAD4;TP53(@>(PTEN1 Chronic(Granulomatous(Disease UCEC 1 1 0.078663977 NCF2,NCF4,CYBA(@>(GMEB2(1.50e@02);NCF2,NCF4,CYBA(@>(ZNF263(1.79e@02);NCF2,CYBA(@>(IRAK1(2.62e@02);NCF2,NCF4,CYBB,CYBA(@>(HELZ2(1.44e@02);NCF2(@>(KRAS(4.41e@02);NCF2,NCF4,CYBB,CYBA(@>(ADAMDEC1(4.66e@02);NCF2,NCF4,CYBB,CYBA(@>(PTEN(1.31e@02);NCF2,NCF4,CYBB,CYBA(@>(PLAGL2(1.18e@02);NCF2,NCF4,CYBB,CYBA(@>(NFATC1(1.48e@02);NCF2,NCF4,CYBB,CYBA(@>(VDR(1.25e@02);NCF2,NCF4,CYBB,CYBA(@>(DNM2(1.64e@02);(@>(HAUS8(4.54e@02);NCF2,NCF4,CYBA(@>(ZGPAT(1.41e@02);NCF4,CYBB,CYBA(@>(NEK8(1.32e@02);NCF2,NCF4,CYBB,CYBA(@>(NFE2L2(1.19e@02);NCF2,NCF4,CYBB,CYBA(@>(CREBBP(1.30e@02);CYBA(@>(TMEM80(2.60e@02);NCF2,NCF4,CYBB,CYBA(@>(CTDP1(1.26e@02);NCF2,NCF4,CYBA(@>(CHMP2A(1.31e@02);CYBA(@>(RPLP2(4.66e@02);NCF2,NCF4,CYBB,CYBA(@>(SAP30BP(1.29e@02);CYBA(@>(POLD4(2.83e@02);NCF2,NCF4,CYBB,CYBA(@>(TPD52L2(1.15e@02);NCF4,CYBB,CYBA(@>(ADAM28(1.19e@02);NCF2,NCF4,CYBB,CYBA(@>(PQLC1(1.18e@02);NCF2,CYBB(@>(CTNNB1(3.35e@02);NCF2,NCF4,CYBB,CYBA(@>(SGK1(1.18e@02);NCF2,NCF4,CYBB,CYBA(@>(MYO9B(1.42e@02);NCF2,CYBB,CYBA(@>(NEU4(1.31e@02);NCF2,NCF4,CYBB,CYBA1 1 Diamond@Blackfan(Anemia UCEC 1 1.14E@06 RPS26,RPS24,RPS10,RPS17,RPS19,RPL5,RPL35A,RPS7,RPL11(@>(Regulation(of(gene(expression(in(beta(cells(RPL14,RPLP2,RPS5,FOXA2,RPL22);RPL11(@>(Validated(targets(of(C@MYC(transcriptional(activation(TAF4B,TERT,TP53,MYC,CREBBP)0.00122045 RPL5,RPS7(@>(NRAS(1.39e@03);RPS19,RPL35A,RPS7(@>(MYC(2.25e@04);RPS19(@>(IRAK1(3.08e@02);RPS26,RPS19,RPS10,RPL35A,RPS7(@>(TRIM28(1.64e@02);RPS26,RPS7(@>(HAUS8(1.64e@03);(@>(CCNE1(6.83e@03);RPS26,RPS7(@>(FTSJ2(1.78e@02);RPS26,RPS19,RPS10,RPL35A,RPS7(@>(NUDT1(2.14e@03);RPS26,RPS24,RPL5,RPS19,RPS10,RPL11,RPL35A,RPS7(@>(RPS5(3.05e@05);RPS24,RPL5,RPS19,RPS10,RPL11,RPL35A(@>(RPLP2(3.96e@04);RPS19,RPS10(@>(POLD4(2.33e@02);RPS26,RPS19,RPS7(@>(TP53(1.72e@03);RPS26,RPS19,RPS7(@>(TACC3(8.31e@04);RPS26,RPS19,RPS7(@>(GEMIN4(2.74e@02);RPS26,RPS10,RPL35A(@>(RNMTL1(3.46e@04);RPS10,RPL35A(@>(ZNF497(1.11e@03);RPS26,RPS19,RPS10,RPL35A,RPS7(@>(TXNL4A(3.23e@02);RPS24,RPL5,RPS19,RPS10,RPL11,RPL35A,RPS7(@>(RPL22(3.92e@05);RPS10(@>(C9orf142(3.77e@03);(@>(TERT(1.23e@02);RPS19(@>(TRAF4(3.89e@02);RPS24,RPL5,RPS19,RPS10,RPL11,RPL35A,RPS7(@>(RPL14(2.97e@04);RPL11,RPL35A,RPS7(@>(RBMX(2.78e@02)0 RPL11(@>(FHIT;RPL35A(@>(MECOM;RPL5(@>(RPLP2;RPS10(@>(RPL14;RPS17(@>(RPS5;RPS19(@>(WWOX,FHIT;RPS24(@>(MECOM;RPS26(@>(RPLP2;RPS7(@>(RPL140.20041667 RPL11(@>(RPL14;RPL5(@>(TP53;RPS10(@>(ESR1;RPS17(@>(RPL22;RPS19(@>(RPS5;RPS24(@>(MYC,RPL14;RPS26(@>(ESR1;RPS7(@>(RPLP2 Inherited(Anomalies(of(the(Skin UCEC 1 TERT 3.29E@08 ATP2A2(@>(nfat(and(hypertrophy(of(the(heart((NFATC1,CREBBP,PIK3R1,PIK3CA);TERT(@>(IL2(signaling(events(mediated(by(PI3K(PIK3CA,MYC,TERT,PIK3R1);TERT(@>(overview(of(telomerase(protein(component(gene(htert(transcriptional(regulation(MYC,TERT,TP53,MZF1,ESR1);TERT(@>(telomeres(telomerase(cellular(aging(and(immortality(MYC,TERT,TP53,RB1,KRAS);KRT1,TERT(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(CTNNB1,CCND1,MYC,TERT);TERT(@>(role(of(nicotinic(acetylcholine(receptors(in(the(regulation(of(apoptosis(PIK3CA,TERT,PIK3R1);TERT(@>(Validated(targets(of(C@MYC(transcriptional(activation(TAF4B,TERT,TP53,MYC,CREBBP);TINF2,TERT,DKC1(@>(Regulation(of(Telomerase(CCND1,MYC,TERT,SIN3A,ESR1)0.183228108 1 1 Combined(Heart(and(Skeletal(Defects UCEC 0.63 CREBBP 2.76E@30 CREBBP(@>(nfat(and(hypertrophy(of(the(heart((NFATC1,CREBBP,PIK3R1,PIK3CA);EP300,CREBBP(@>(IFN@gamma(pathway(PIK3CA,CREBBP,PIK3R1);EP300,CREBBP(@>(Direct(p53(effectors(VDR,TP53,PIDD,RB1,PTEN,CREBBP);EP300,CREBBP(@>(FOXA1(transcription(factor(network(FOXA2,CREBBP,NKX3@1,ESR1);EP300,CREBBP(@>(transcription(regulation(by(methyltransferase(of(carm1(CREBBP,PRKAR1B);EP300(@>(cell(cycle:(g2/m(checkpoint(TP53,MYT1);EP300,CREBBP(@>(carm1(and(regulation(of(the(estrogen(receptor(CREBBP,ESR1);CREBBP(@>(wnt(signaling(pathway(CTNNB1,CCND1,MYC,CREBBP);EP300(@>(Validated(nuclear(estrogen(receptor(alpha(network(CCND1,MYC,UBE2M,ESR1);EP300(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(CTNNB1,CCND1,MYC,TERT);EP300,CREBBP(@>(mechanism(of(gene(regulation(by(peroxisome(proliferators(via(ppara(MYC,PRKAR1B,RB1,CREBBP);CREBBP(@>(regulation(of(transcriptional(activity(by(pml(TP53,CREBBP,RB1);EP300,CREBBP(@>(E2F(transcription(factor(network(MYC,CCNE1,TRIM28,CREBBP,RB1);EP300,CREBBP(@>(il@7(signal(transduc1 0.33848148 1 Hereditary(Sensory(Neuropathy UCEC 0.71 NEFL,DNM2 0.096035489 NTRK1(@>(Trk(receptor(signaling(mediated(by(PI3K(and(PLC@gamma(CCND1,PIK3CA,NRAS,PIK3R1,KRAS);NDRG1(@>(Direct(p53(effectors(VDR,TP53,PIDD,RB1,PTEN,CREBBP);NTRK1(@>(trka(receptor(signaling(pathway(PIK3CA,PIK3R1);NTRK1(@>(p73(transcription(factor(network(MYC,RNF43,RB1,WWOX);NTRK1(@>(p75(NTR)@mediated(signaling(PIK3CA,IRAK1,OMG,TP53,PIK3R1);DNM2(@>(PAR1@mediated(thrombin(signaling(events(PIK3CA,GNAQ,PIK3R1,DNM2);PMP22(@>(a6b1(and(a6b4(Integrin(signaling(ERBB2,PIK3CA,PIK3R1,ERBB3)0.130354569 1 1 Li(Fraumeni(and(Related(Syndromes UCEC 0.63 TP53 1.96E@18 TP53(@>(chaperones(modulate(interferon(signaling(pathway(TP53,RB1);TP53(@>(Direct(p53(effectors(VDR,TP53,PIDD,RB1,PTEN,CREBBP);TP53(@>(LKB1(signaling(events(MYC,TP53,ESR1);TP53,CHEK2(@>(cell(cycle:(g2/m(checkpoint(TP53,MYT1);TP53(@>(estrogen(responsive(protein(efp(controls(cell(cycle(and(breast(tumors(growth(TP53,ESR1);TP53(@>(overview(of(telomerase(protein(component(gene(htert(transcriptional(regulation(MYC,TERT,TP53,MZF1,ESR1);CDKN2A(@>(Coregulation(of(Androgen(receptor(activity(CTNNB1,CCND1,CASP8,NKX3@1);CDKN2A,TP53(@>(Hypoxic(and(oxygen(homeostasis(regulation(of(HIF@1@alpha(TP53,ARNT);TP53(@>(telomeres(telomerase(cellular(aging(and(immortality(MYC,TERT,TP53,RB1,KRAS);CDKN2A(@>(Regulation(of(nuclear(beta(catenin(signaling(and(target(gene(transcription(CTNNB1,CCND1,MYC,TERT);TP53(@>(btg(family(proteins(and(cell(cycle(regulation(CCND1,TP53,RB1);TP53(@>(Transcriptional((activation(of((cell(cycle(inhibitor(p21(TP53);TP53,CHEK2(@>(PLK3(signaling(events(CCNE1,TP53);TP53(@>(p53(signaling(pathway(CCND1,CCNE1,TP53,0.458169296 0.57574242 1 Lipoprotein(Deficiencies UCEC 1 1 0.064216795 MTTP,APOB,LCAT,SAR1B,APOA1(@>(A1BG(1.19e@02);MTTP,APOB,LCAT,SAR1B,APOA1(@>(SLC27A5(7.36e@03);MTTP,APOB,LCAT,SAR1B,APOA1(@>(HPD(1.19e@02);MTTP,APOB,LCAT,SAR1B,APOA1(@>(ATRN(2.26e@02);MTTP,APOB,LCAT,SAR1B,APOA1,ABCA1(@>(NEU4(1.39e@02)1 1

Supplementary Table 4: Continuation of Supplementary Table 3

MD C gene_enrichmentgeneIntersection pathway_correlationpathway coex_CG coexpression humannet_sethumannet biogrid_set biogrid ! Chronic!Granulomatous!Disease LUAD 1 0.384190056 0.078679242 NCF4,CYBA!E>!CCND3(3.54eE02);NCF2,NCF4,CYBA!E>!ZGPAT(1.74eE02);NCF2,NCF4,CYBB,CYBA!E>!RIT1(1.48eE02);NCF2,NCF4,CYBB,CYBA!E>!ITGAX(1.79eE02);NCF2,NCF4,CYBB,CYBA!E>!DNAJC5(1.82eE02);NCF2,NCF4,CYBB,CYBA!E>!PTGER4(1.60eE02);NCF2,NCF4,CYBB,CYBA!E>!SIRPB1(1.30eE02);NCF2,NCF4,CYBB,CYBA!E>!TNFSF13B(4.88eE02);NCF2,NCF4,CYBB,CYBA!E>!PQLC1(1.33eE02);NCF2,NCF4,CYBB,CYBA!E>!TBL1X(1.72eE02);NCF4,CYBA!E>!ARHGEF6(2.95eE02);NCF2,NCF4,CYBB,CYBA!E>!GNG2(1.37eE02);NCF2,NCF4,CYBB,CYBA!E>!NFATC1(1.81eE02);NCF2,NCF4!E>!U2AF1(3.61eE02);NCF2,NCF4,CYBB,CYBA!E>!BTK(1.94eE02);NCF2,NCF4,CYBB,CYBA!E>!SAMSN1(1.35eE02);NCF2,NCF4,CYBB,CYBA!E>!LPAR6(2.19eE02);NCF2,NCF4!E>!IL18RAP(1.83eE02);NCF2,NCF4,CYBB,CYBA!E>!RB1(3.72eE02);NCF2,NCF4,CYBB!E>!ADNP2(3.36eE02);NCF2,NCF4,CYBB,CYBA!E>!MFSD7(3.07eE02);NCF2,NCF4,CYBB,CYBA!E>!PMAIP1(1.26eE02);NCF4,CYBB,CYBA!E>!TBX21(1.81eE02);NCF2,NCF4,CYBB,CYBA!E>!ANKRD44(1.82eE02);NCF2,NCF4,CYBB,CYBA!E>!AQP9(1.28eE02);NCF2,NCF4,CYBB,CYBA!E>!PPM1F(1.53eE02);NCF2,NCF4,CYBB,CYBA!E>!CTDP1(1.37eE02);NCF2,NCF4,CYBB,CYB1 1 ! Congenital!Ichthyosis LUAD 1 1 0.020062495 ALOX12B,SPINK5,CSTA,TGM1!E>!SPRR3(8.05eE03);ALOX12B!E>!KRT28(3.04eE03);!E>!FLG(1.68eE03);ALOX12B,SPINK5,CSTA,ABCA12,TGM1!E>!SERPINB13(1.68eE03);ALOX12B,SPINK5,KRT2,ABCA12!E>!POF1B(8.05eE03)1 1 ! Disorders!of!Phosphorous!Metabolism LUAD 1 CYP27B1 0.003900738 FGF23!E>!SyndecanE2Emediated!signaling!events(HRAS,LAMA3,FGF19,NF1);CYP27B1!E>!Vitamin!D!(calciferol)!metabolism(GC,CYP27B1);FGF23!E>!SyndecanE1Emediated!signaling!events(COL11A1,COL5A2,MET,FGF19,COL5A1,COL3A1)1 1 1 ! DiamondEBlackfan!Anemia LUAD 1 1 0.006070507 RPS26,RPS19,RPS10,RPS7!E>!BYSL(4.40eE02);RPL5,RPS10,RPL11!E>!CCND3(3.53eE02);!E>!UTF1(8.74eE03);RPS24,RPL5!E>!ALG10(3.42eE02);RPS26,RPS19,RPS7!E>!GEMIN4(2.58eE02);RPS26!E>!CDKN2A(9.01eE03);RPS26,RPL35A,RPS7!E>!U2AF1(4.63eE04);RPS26,RPS7!E>!FANCD2(1.47eE02);RPS26,RPS19,RPS10,RPL35A!E>!LAGE3(4.10eE02);!E>!CTCFL(1.14eE02);!E>!MDM2(4.11eE02);!E>!VENTX(7.84eE03);RPS26,RPS19,RPS7!E>!TP53(1.67eE03);RPS26,RPS19,RPS10,RPS7!E>!EIF4EBP1(1.12eE02);RPS26,RPS19,RPS7!E>!TFDP1(2.13eE03);RPS26,RPS19,RPS10,RPL35A,RPS7!E>!TXNL4A(3.14eE02);!E>!TERT(1.10eE02);RPS19,RPS7!E>!CNIH1(4.86eE04);RPL5,RPS7!E>!NRAS(1.31eE03);RPS26,RPS10,RPL35A!E>!RNMTL1(3.54eE04);RPS26,RPS7!E>!METTL1(2.41eE02);RPS26,RPS19,RPS10,RPS7!E>!HAX1(1.49eE02)1 1 ! Inherited!Anomalies!of!the!Skin LUAD 1 TERT 3.07EE07 TERT!E>!overview!of!telomerase!protein!component!gene!htert!transcriptional!regulation(TERT,TP53);TERT!E>!telomeres!telomerase!cellular!aging!and!immortality(TERT,TP53,RB1,KRAS);KRT1,TERT!E>!Regulation!of!nuclear!beta!catenin!signaling!and!target!gene!transcription(CDKN2A,TBL1X,CCND1,APC,TERT,SMARCA4)0.086097896 KRT6A,KRT16!E>!CYP27B1(1.45eE02);WRAP53,TERC,DKC1,NHP2,NOP10!E>!GEMIN4(4.63eE02);TERC,NHP2,NOP10!E>!FAM58A(4.53eE02);KRT6A,TERC,NHP2,KRT16,NOP10!E>!HRAS(4.57eE02);TERC,NHP2,NOP10!E>!SLC10A3(4.68eE02);KRT6A,NHP2,KRT16,NOP10!E>!NXN(4.37eE02);!E>!AKR1B10(4.64eE02)1 1 ! Spinocerebellar!Ataxia LUAD 1 ATM 8.60EE07 PRKCG!E>!EGFR!Inhibitor!Pathway,!Pharmacodynamics(ERBB2,NRAS,HRAS,EGFR,KRAS);ATM!E>!apoptotic!signaling!in!response!to!dna!damage(ATM,TP53);ATM!E>!p53!pathway(CDKN2A,ATM,TP53,MDM2);ATM!E>!ATM!pathway(MDM2,ATM,FANCD2);ATM!E>!cell!cycle:!g2/m!checkpoint(MDM2,ATM,TP53,MYT1);ATM!E>!ATM!mediated!response!to!DNA!doubleEstrand!break(ATM);ATM!E>!cdc25!and!chk1!regulatory!pathway!in!response!to!dna!damage(ATM,MYT1);ATM!E>!role!of!brca1!brca2!and!atr!in!cancer!susceptibility(ATM,TP53,FANCD2);ATM!E>!BARD1!signaling!events(ATM,TP53,FANCD2);ATM!E>!atm!signaling!pathway(MDM2,ATM,TP53);PRKCG!E>!IL8E!and!CXCR2Emediated!signaling!events(PLCB1,GNG2,HCK);ATM!E>!rb!tumor!suppressor/checkpoint!signaling!in!response!to!dna!damage(ATM,TP53,RB1,MYT1);PRKCG!E>!IL8E!and!CXCR1Emediated!signaling!events(PLCB1,GNG2,HCK);ATM!E>!Metformin!Pathway,!Pharmacodynamic(ATM,STK11,PRKAA1);ATM!E>!E2F!transcription!factor!network(TFDP1,CCND3,ATM,CDKN2A,RB1);ATM!E>!hypoxia!and!p53!in!the!cardiovascular!system(MDM2,ATM,TP53);TBP!E>!Glucocorticoid!rece0.001274155 SYT14!E>!SCG2(2.09eE02);APTX,ZNF592,ATXN2,TTBK2,TBP,KCNC3,ITPR1,NOP56,SETX,SYT14,C10orf2!E>!KIAA0907(2.58eE03);JPH3,TDP1,KCNC3,FGF14!E>!SAMD10(1.63eE02);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B!E>!DOC2B(3.57eE03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B!E>!C1orf173(3.57eE03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B!E>!TMEM132D(4.61eE03);JPH3,APTX,ZNF592,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,TBP,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B!E>!NF1(2.40eE03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B!E>!OPCML(2.50eE03);JPH3,ZNF592,SPTBN2,TTBK2,KCNC3,PRNP,PRKCG,FGF14,PPP2R2B,ATXN1!E>!DNAJC5(3.21eE02);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,TBP,KCNC3,PRNP,ITPR1,PRKCG,FGF14,SETX,SYT14,PPP2R2B!E>!TTC33(2.02eE03);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,TBP,KCNC3,ATXN10,ITPR1,PRKCG,FGF14,AFG3L2,SYT14,PPP2R2 GlucoseE6EPhosphate!Dehydrogenase!DeficiencyLUAD 0.05 UBL4A,G6PD 0.034828736 NA!E>!NA(NA) E1 1 1 ! Hypopituitarism LUAD 1 BTK 0.034828736 FGFR1!E>!SyndecanE2Emediated!signaling!events(HRAS,LAMA3,FGF19,NF1);BTK!E>!bcr!signaling!pathway(NFATC1,HRAS,BTK,PPP3CA);BTK!E>!EPO!signaling!pathway(HRAS,PTPN11,BTK);POU1F1!E>!Glucocorticoid!receptor!regulatory!network(NFATC1,TP53,TBX21,MDM2,SMARCA4,GATA3);GH1!E>!trefoil!factors!initiate!mucosal!healing(ERBB2,HRAS,EGFR);FGFR1!E>!SyndecanE1Emediated!signaling!events(COL11A1,COL5A2,MET,FGF19,COL5A1,COL3A1)1 0.2886 1 ! Combined!Heart!and!Skeletal!Defects LUAD 1 3.78EE08 EP300,CREBBP!E>!Direct!p53!effectors(BCL2L14,PMAIP1,TP53,EGFR,RB1,MET,APC,MDM2,SMARCA4,TFDP1);EP300,CREBBP!E>!p53!pathway(CDKN2A,ATM,TP53,MDM2);EP300!E>!cell!cycle:!g2/m!checkpoint(MDM2,ATM,TP53,MYT1);EP300!E>!Validated!transcriptional!targets!of!AP1!family!members!Fra1!and!Fra2(CCND1,NFATC1,LAMA3,CDKN2A);EP300!E>!Validated!transcriptional!targets!of!TAp63!isoforms(MDM2,CDKN2A,PMAIP1,SPATA18);EP300!E>!Regulation!of!nuclear!beta!catenin!signaling!and!target!gene!transcription(CDKN2A,TBL1X,CCND1,APC,TERT,SMARCA4);CREBBP!E>!regulation!of!transcriptional!activity!by!pml(TP53,RB1);EP300,CREBBP!E>!E2F!transcription!factor!network(TFDP1,CCND3,ATM,CDKN2A,RB1);EP300!E>!p73!transcription!factor!network(JAG2,MDM2,NTRK1,RB1,WWOX);EP300!E>!hypoxia!and!p53!in!the!cardiovascular!system(MDM2,ATM,TP53);EP300,CREBBP!E>!Glucocorticoid!receptor!regulatory!network(NFATC1,TP53,TBX21,MDM2,SMARCA4,GATA3);EP300!E>!melanocyte!development!and!pigmentation!pathway(HRAS,KIT)0.71670305 1 1 ! Neurofibromatosis LUAD 0.66 NF1 0.001098828 NF1!E>!Regulation!of!Ras!family!activation(HRAS,NRAS,NF1,KRAS);NF1!E>!SyndecanE2Emediated!signaling!events(HRAS,LAMA3,FGF19,NF1);NF1!E>!chromatin!remodeling!by!hswi/snf!atpEdependent!complexes(ARID1A,SMARCA4,NF1)0.675271975 0.22904762 NF2!E>!CDKN2A 0.60125 ! Hereditary!Sensory!Neuropathy LUAD 1 NTRK1 1.95EE05 NTRK1!E>!Trk!receptor!signaling!mediated!by!PI3K!and!PLCEgamma(CCND1,HRAS,NRAS,NTRK1,KRAS);NTRK1!E>!ARMSEmediated!activation(NTRK1,BRAF);NDRG1!E>!Direct!p53!effectors(BCL2L14,PMAIP1,TP53,EGFR,RB1,MET,APC,MDM2,SMARCA4,TFDP1);EGR2!E>!Validated!transcriptional!targets!of!TAp63!isoforms(MDM2,CDKN2A,PMAIP1,SPATA18);NTRK1!E>!TRKA!activation!by!NGF(NTRK1);NTRK1!E>!NGF!signalling!via!TRKA!from!the!plasma!membrane(NTRK1);NTRK1!E>!Signalling!to!ERKs(NTRK1);NTRK1!E>!Signalling!to!STAT3(NTRK1);NTRK1!E>!trka!receptor!signaling!pathway(HRAS,NTRK1);RAB7A!E>!IL8E!and!CXCR2Emediated!signaling!events(PLCB1,GNG2,HCK);NTRK1!E>!Frs2Emediated!activation(NTRK1,BRAF);NTRK1!E>!Signalling!to!p38!via!RIT!and!RIN(NTRK1,BRAF);NTRK1!E>!p73!transcription!factor!network(JAG2,MDM2,NTRK1,RB1,WWOX);PMP22!E>!a6b1!and!a6b4!Integrin!signaling(ERBB2,MET,HRAS,LAMA3,EGFR)0.128961218 1 1 ! Severe!Combined!Immunodeficiency LUAD 1 0.009297242 DCLRE1C!E>!ATM!pathway(MDM2,ATM,FANCD2);ADA!E>!Validated!transcriptional!targets!of!TAp63!isoforms(MDM2,CDKN2A,PMAIP1,SPATA18);ADA!E>!p73!transcription!factor!network(JAG2,MDM2,NTRK1,RB1,WWOX);JAK3,IL2RG!E>!IL2Emediated!signaling!events(HRAS,NRAS,PTPN11,KRAS);JAK3!E>!il!6!signaling!pathway(PTPN11,HRAS);ADA!E>!Validated!transcriptional!targets!of!deltaNp63!isoforms(COL5A1,CDKN2A,ATM,MDM2)0.001684618 CIITA,RFX5,DCLRE1C!E>!C11orf35(2.10eE02);!E>!GATA3(2.09eE02);ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C,RAG2,RAG1!E>!CCND3(1.40eE04);ZAP70,IL2RG,JAK3,ADA,PTPRC,IL7R,CD3D!E>!ZGPAT(2.99eE03);ZAP70,IL2RG,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,CD3D!E>!PTGER4(1.27eE02);IL2RG,CIITA,JAK3,ADA,PNP,PTPRC!E>!TNFSF13B(4.44eE02);RFXANK!E>!RBM10(2.42eE03);ZAP70,IL2RG,RFXAP,RFX5,AK2,IL7R,CD3D,DCLRE1C!E>!ALG10(7.50eE04);IL2RG,RFX5,DCLRE1C!E>!CMTR2(2.45eE03);IL2RG,JAK3,PNP,RFX5,PTPRC,DCLRE1C!E>!AOAH(2.05eE02);IL2RG,JAK3,PTPRC!E>!TBL1X(2.39eE02);ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!ARHGEF6(9.58eE04);ZAP70,IL2RG,CIITA,JAK3,RFXAP,RFX5,PTPRC,CD3D,DCLRE1C!E>!ARID1A(5.42eE03);ZAP70,IL2RG,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,DCLRE1C!E>!GNG2(3.39eE02);IL2RG,JAK3,ADA,PNP,RFX5,PTPRC!E>!NFATC1(1.94eE02);ADA,PNP,AK2!E>!U2AF1(9.03eE04);ZAP70,IL2RG,JAK3,RFXAP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C,RAG2,RAG1!E>!ARID2(1.03eE02);IL2RG,CIITA,JAK3,RFX5,PTPRC,DCLRE1C!E>!BTK(3.08eE03);ZAP70,IL2RG,JAK3,PTPRC,IL7R,CD3D,DCLRE1C,RAG2,RAG1!E>!THEMIS(2.35eE04);RFXA1 1 ! Specified!Hamartoses LUAD 0.97 STK11 2.59EE05 VHL!E>!vegf!hypoxia!and!angiogenesis(HRAS,KDR,ARNT);PTEN!E>!Direct!p53!effectors(BCL2L14,PMAIP1,TP53,EGFR,RB1,MET,APC,MDM2,SMARCA4,TFDP1);VHL!E>!Hypoxic!and!oxygen!homeostasis!regulation!of!HIFE1Ealpha(CDKN2A,TP53,ARNT);STK11!E>!Metformin!Pathway,!Pharmacodynamic(ATM,STK11,PRKAA1)0.717635623 0.25727907 1 ! Li!Fraumeni!and!Related!Syndromes LUAD 0.09 CDKN2A,TP53 4.67EE28 TP53!E>!chaperones!modulate!interferon!signaling!pathway(TP53,RB1);TP53!E>!apoptotic!signaling!in!response!to!dna!damage(ATM,TP53);TP53!E>!Direct!p53!effectors(BCL2L14,PMAIP1,TP53,EGFR,RB1,MET,APC,MDM2,SMARCA4,TFDP1);TP53,CDKN2A,CHEK2!E>!p53!pathway(CDKN2A,ATM,TP53,MDM2);CHEK2!E>!ATM!pathway(MDM2,ATM,FANCD2);TP53,CHEK2!E>!cell!cycle:!g2/m!checkpoint(MDM2,ATM,TP53,MYT1);CDKN2A!E>!Validated!transcriptional!targets!of!AP1!family!members!Fra1!and!Fra2(CCND1,NFATC1,LAMA3,CDKN2A);CDKN2A!E>!Validated!transcriptional!targets!of!TAp63!isoforms(MDM2,CDKN2A,PMAIP1,SPATA18);TP53!E>!overview!of!telomerase!protein!component!gene!htert!transcriptional!regulation(TERT,TP53);TP53,CHEK2!E>!role!of!brca1!brca2!and!atr!in!cancer!susceptibility(ATM,TP53,FANCD2);CDKN2A,TP53!E>!Hypoxic!and!oxygen!homeostasis!regulation!of!HIFE1Ealpha(CDKN2A,TP53,ARNT);TP53!E>!telomeres!telomerase!cellular!aging!and!immortality(TERT,TP53,RB1,KRAS);CDKN2A!E>!Regulation!of!nuclear!beta!catenin!signaling!and!target!gene!transcription(CDKN2A,TBL1X,CCND10.719639888 0 CDKN2A!E>!MDM2;CHEK2!E>!CCND3,ATM;TP53!E>!TP53,SMAD41 ! Genetic!Anomalies!of!Leukocytes LUAD 1 0.02688935 ITGB2!E>!Beta2!integrin!cell!surface!interactions(ITGAX,SPON2,FGB)0.471117151 1 1 ! Lipoprotein!Deficiencies LUAD 1 MTTP 1 0.097427692 APOB,LCAT,SAR1B,APOA1!E>!SLC26A1(2.83eE02);APOB,LCAT,SAR1B,APOA1!E>!PCK1(3.92eE02);APOB,LCAT,SAR1B,APOA1!E>!PROS1(1.82eE02);!E>!AKR1C2(2.10eE02);APOB,LCAT,SAR1B,APOA1!E>!EHHADH(1.83eE02);APOB,LCAT,SAR1B,APOA1!E>!BHMT(1.95eE02);APOB,LCAT,SAR1B,APOA1!E>!GBA3(1.93eE02);APOB,LCAT,SAR1B,APOA1!E>!ABCG5(2.40eE02);APOB,LCAT,SAR1B,APOA1!E>!MTTP(2.00eE02);!E>!CD5L(2.00eE02)1 1 ! Disorders!of!Urea!Cycle!Metabolism LUAD 1 1 0.069332922 ASS1,NAGS,ARG1,ASL,CPS1!E>!GC(1.50eE02);ASS1,NAGS,ARG1,ASL,CPS1!E>!SLC26A1(1.99eE02);ASS1,NAGS,ARG1,ASL,CPS1!E>!FGB(2.95eE02);ASS1,NAGS,ARG1,ASL,CPS1!E>!PROS1(1.02eE02);ASS1,NAGS,ASL,CPS1!E>!HSBP1L1(9.88eE03);ASS1!E>!AKR1C2(3.55eE02);ASS1,NAGS,ARG1,ASL,CPS1!E>!EHHADH(1.20eE02);NAGS,ARG1,ASL,CPS1!E>!SOWAHB(1.38eE02);ASS1,NAGS,ARG1,ASL,CPS1!E>!CYP4V2(3.55eE02);ASS1,NAGS,ARG1,ASL,CPS1!E>!GBA3(8.52eE03);ASS1,NAGS,ARG1,ASL,CPS1!E>!ABCG5(9.51eE03);ASS1,NAGS,ARG1,ASL,CPS1!E>!MTTP(9.88eE03);!E>!CD5L(9.22eE03)1 1 ! Retinitis!Pigmentosa LUAD 1 PDE6B 0.828420314 6.10EE15 RPGR,CRX,SNRNP200,CA4,EYS,CRB1,CERKL,PRPF3,TULP1,C2orf71,TOPORS,FAM161A!E>!KIAA0907(3.68eE02);KLHL7,SPATA7,CRB1,MERTK,CERKL,FAM161A!E>!DOC2B(3.41eE02);!E>!MUC16(2.78eE04);SNRNP200,CRB1,PRPF31,CERKL,IDH3B,PRPF8,FAM161A!E>!UCKL1(2.83eE03);CRX,FSCN2,RDH12,PRPH2,CNGB1,EYS,CRB1,CERKL,TULP1,ROM1,PRCD,IMPG2,C2orf71,RBP3!E>!GPR112(5.36eE07);KLHL7,SPATA7,CRB1,PRCD,IMPG2,C2orf71,FAM161A!E>!TMEM132D(3.43eE02);SNRNP200!E>!FZD10(1.68eE02);IMPDH1!E>!CYP27B1(1.33eE02);CRX,SNRNP200,KLHL7,EYS,SPATA7,CRB1,CERKL,USH2A,PRCD,IMPG2,C2orf71,RBP3,FAM161A!E>!GTF2I(4.98eE02);!E>!SLC22A6(1.68eE02);CNGA1!E>!ANKRD37(1.67eE02);CRX,FSCN2,PRPH2,CNGB1,KLHL7,EYS,SPATA7,CRB1,MERTK,CERKL,TULP1,ROM1,PRCD,IMPG2,C2orf71,RBP3,FAM161A!E>!RIMS2(3.67eE04);SPATA7,CRB1,PRPF3!E>!ITGB8(1.07eE02);RPGR,CA4,IMPDH1,PRPF3,BEST1,RP2,TOPORS,SEMA4A!E>!TBL1X(2.76eE02);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,EYS,CRB1,NRL,PDE6A,PDE6G,CERKL,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,PROM1,CNGA1,RP1,PRCD,IMPG2,C2orf71,RPE65,SAG,RBP3,FAM161A,ABCA4!E>!RP1L1(1.27eE17);CRX,FSCN0.02531579 CNGA1!E>!CNGA2,LRRC32,EIF4G3;CNGB1!E>!PABPC3,PDE6B,PDE6B,TXNL4A;IMPDH1!E>!PRPF6,PABPC3;PDE6A!E>!TXNL4A;PDE6G!E>!PRPF6,EIF4G3;PRPF3!E>!PABPC3;PRPF31!E>!TXNL4A;PRPF8!E>!PRPF6,U2AF1,U2AF1;RP9!E>!PABPC3;SNRNP200!E>!TXNL4A1 ! Haemophilia LUAD 0.7 F8 0.071266219 VWF!E>!Platelet!Aggregation!Inhibitor!Pathway,!Pharmacodynamics(COL3A1,COL4A2,FGB,PLCB1);F8,F9!E>!intrinsic!prothrombin!activation!pathway(F8,PROS1,COL4A2,FGB)1 0.7955 0.52680952 ! Chronic!Granulomatous!Disease LUSC 1 1 0.067971546 NCF2,NCF4,CYBB,CYBA!E>!B2M(1.31eE02);NCF2,NCF4,CYBB,CYBA!E>!MARCH1(8.59eE03);NCF2,NCF4,CYBB!E>!USP25(1.12eE02);NCF2,NCF4,CYBB,CYBA!E>!NFE2L2(8.87eE03);NCF2,NCF4,CYBB,CYBA!E>!LYZ(2.21eE02);NCF2,NCF4,CYBB,CYBA!E>!PTEN(1.06eE02);NCF2,NCF4,CYBB,CYBA!E>!HDAC10(1.05eE02);NCF2,NCF4,CYBB,CYBA!E>!KDM5A(1.96eE02);NCF2,NCF4,CYBB,CYBA!E>!NFATC1(1.20eE02);NCF4,CYBB,CYBA!E>!CHKB(2.32eE02);NCF2,NCF4,CYBB,CYBA!E>!TRABD(1.01eE02);NCF2,NCF4,CYBB,CYBA!E>!CREBBP(1.06eE02);NCF2,NCF4,CYBB,CYBA!E>!ODF3B(1.41eE02);NCF2,NCF4,CYBB,CYBA!E>!CTDP1(9.25eE03);NCF2,NCF4,CYBB,CYBA!E>!REL(9.24eE03);NCF2,NCF4,CYBB,CYBA!E>!KDM6A(1.20eE02);NCF2,NCF4,CYBB,CYBA!E>!METRNL(9.24eE03);NCF2,NCF4,CYBB,CYBA!E>!PQLC1(8.59eE03);NCF2,NCF4,CYBB,CYBA!E>!LPAR6(1.21eE02);NCF2,NCF4,CYBB,CYBA!E>!BID(1.21eE02);NCF2,NCF4,CYBB,CYBA!E>!RB1(1.92eE02);NCF2,NCF4,CYBB,CYBA!E>!PIM3(8.92eE03);NCF4!E>!NINJ2(1.01eE02);NCF2,NCF4,CYBB,CYBA!E>!EVI2A(1.71eE02);NCF2,NCF4,CYBB,CYBA!E>!EVI2B(9.25eE03);NCF2,NCF4!E>!EXOC3(2.80eE02);NCF2,NCF4,CYBB,CYBA!E>!NOTCH1(8.21eE03);NCF2,NCF4,CY1 1 ! Congenital!Ichthyosis LUSC 1 1 0.006200982 ALOX12B,ALOXE3,SPINK5,CSTA,KRT2,ABCA12,TGM1!E>!CERS3(3.87eE04);CSTA,NIPAL4,LIPN,ABHD5!E>!NFE2L2(3.14eE02);ALOX12B,SPINK5,CSTA,TGM1!E>!PAX9(3.14eE02);ALOX12B,ALOXE3,CSTA,TGM1!E>!PARD6G(4.70eE02);CSTA,NIPAL4,LIPN,ABHD5!E>!NOTCH1(2.92eE02);ABCA12,TGM1!E>!EGFR(3.98eE02)0.79744737 1 ! DiamondEBlackfan!Anemia LUSC 1 0.959794083 0.002784462 RPS26!E>!CDKN2A(9.26eE03);!E>!YEATS4(1.30eE03);RPS26,RPS19,RPS7!E>!TP53(1.51eE03);RPS26,RPS19,RPL35A,RPS7!E>!PDCD6(1.33eE04);RPS26,RPS19,RPS10,RPL35A,RPS7!E>!TXNL4A(3.08eE02);!E>!NMU(2.86eE02);RPL5,RPS7!E>!TTF2(4.93eE02);RPS26,RPS19,RPS7!E>!TYMS(2.02eE03);RPS26,RPS7!E>!TRIP13(9.59eE03)1 1 ! Spinocerebellar!Ataxia LUSC 1 0.977629639 0.01980021 JPH3,CACNA1A,POLG,ATXN2,SYNE1,SPTBN2,TTBK2,KCNC3,PRKCG,FGF14,PPP2R2B!E>!SBF1(9.84eE03);JPH3,ZNF592,ATM,TTBK2,KCNC3,ITPR1,SETX,PPP2R2B!E>!MARCH1(3.18eE02);JPH3,CACNA1A,SPTBN2,TTBK2,KCNC3,ATXN10,PRKCG,FGF14,SYT14,PPP2R2B!E>!L1CAM(1.18eE02);JPH3,ZNF592,CACNA1A,ATXN2,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B,ATXN1!E>!CCSER1(2.45eE03);JPH3,ZNF592,CACNA1A,ATXN2,PDYN,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,AFG3L2,SYT14,PPP2R2B!E>!PARK2(2.80eE03);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B!E>!MAPK8IP2(2.90eE03);TDP1,ATM,TBP,NOP56,SYT14,C10orf2!E>!CCDC77(1.54eE02);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B!E>!CLCN4(2.99eE03);POLG,ATXN2,TBP,NOP56,AFG3L2,C10orf2!E>!BRD9(2.95eE03);ZNF592,POLG,ATXN7,ATXN2,TDP1,SYNE1,ATM,TTBK2,TBP,ITPR1,SETX,ATXN1!E>!BRD1(3.18eE02);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B!E>!CSMD3(2.99eE03);JPH3,CACNA1A1 0.73607576 ! Combined!Heart!and!Skeletal!Defects LUSC 0.6 CREBBP 3.46EE14 CREBBP!E>!nfat!and!hypertrophy!of!the!heart!(NFATC1,CREBBP,PIK3CA);CREBBP!E>!inhibition!of!huntingtons!disease!neurodegeneration!by!histone!deacetylase!inhibitors(CREBBP);EP300,CREBBP!E>!Direct!p53!effectors(TP53,BID,EGFR,RB1,PTEN,CREBBP);EP300,CREBBP!E>!p53!pathway(CDKN2A,TP53,CREBBP);CREBBP!E>!NotchEHLH!transcription!pathway(CREBBP);EP300,CREBBP!E>!mechanism!of!gene!regulation!by!peroxisome!proliferators!via!ppara(FAT1,RB1,CREBBP);CREBBP!E>!regulation!of!transcriptional!activity!by!pml(TP53,CREBBP,RB1);EP300,CREBBP!E>!E2F!transcription!factor!network(TYMS,CDKN2A,CREBBP,RB1);EP300,CREBBP!E>!ilE7!signal!transduction(PIK3CA,CREBBP);EP300!E>!p73!transcription!factor!network(MAPK11,CDK6,RB1,WWOX);EP300,CREBBP!E>!Glucocorticoid!receptor!regulatory!network(NFATC1,TP53,MAPK11,CREBBP);CREBBP!E>!Signaling!events!mediated!by!Stem!cell!factor!receptor!(cEKit)(PIK3CA,CREBBP,PTEN);EP300!E>!ATFE2!transcription!factor!network(MAPK11,RB1,NF1);CREBBP!E>!Signaling!events!mediated!by!TCPTP(PIK3CA,EGFR,CREBBP);EP300,CREBBP!E>!F1 0.0925 CREBBP!E>!TP53,TP53;EP300!E>!CREBBP1 ! Neurofibromatosis LUSC 0.53 NF1 0.236599603 0.719411434 0 NF1!E>!EVI2A,CDKN2A;NF2!E>!NF11 ! Severe!Combined!Immunodeficiency LUSC 1 0.23074305 0.001138137 ZAP70,IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!B2M(6.47eE03);IL2RG,CIITA,JAK3,ADA,PNP,RFX5,PTPRC,DCLRE1C!E>!PLEKHO1(2.51eE02);ZAP70,IL2RG,CIITA,JAK3,PTPRC,IL7R,CD3D,DCLRE1C!E>!BRD1(1.02eE02);!E>!YEATS4(1.46eE02);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!HDAC10(2.23eE04);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!KDM5A(8.24eE03);IL2RG,JAK3,ADA,PNP,RFX5,PTPRC!E>!NFATC1(1.79eE02);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!FOXP1(2.39eE02);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!CHKB(1.16eE03);ZAP70,IL2RG,CIITA,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!TRABD(1.29eE04);RFXAP,RFX5,DCLRE1C!E>!PRDM15(8.47eE03);ZAP70,IL2RG,JAK3,PNP,RFX5,PTPRC,IL7R,DCLRE1C!E>!CREBBP(3.99eE02);ZAP70,IL2RG,CIITA,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!CTDP1(3.26eE03);IL2RG,CIITA,JAK3,ADA,PNP,RFX5,PTPRC!E>!REL(3.66eE02);ZAP70,IL2RG,CIITA,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!KDM6A(2.09eE02);ZAP70,IL2RG,RFX5,DCLRE1C!E>!ZBED4(2.61eE05);ADA,AK2,DCLRE1C!E>!CDK6(2.85eE04);ZA1 1 ! Specified!Hamartoses LUSC 0.63 PTEN 8.39EE06 PTEN!E>!skeletal!muscle!hypertrophy!is!regulated!via!aktEmtor!pathway(PIK3CA,PTEN);PTEN!E>!regulation!of!eifE4e!and!p70s6!kinase(PIK3CA,PTEN);PTEN!E>!mtor!signaling!pathway(PTEN,PIK3CA);PTEN!E>!Direct!p53!effectors(TP53,BID,EGFR,RB1,PTEN,CREBBP);VHL!E>!Hypoxic!and!oxygen!homeostasis!regulation!of!HIFE1Ealpha(CDKN2A,TP53);PTEN!E>!RhoA!signaling!pathway(PTEN,MAPK12,SLC9A3);PTEN!E>!pten!dependent!cell!cycle!arrest!and!apoptosis(PTEN,PIK3CA);PTEN!E>!Signaling!events!mediated!by!Stem!cell!factor!receptor!(cEKit)(PIK3CA,CREBBP,PTEN)0.619582187 0 PTEN!E>!TTF2;SDHB!E>!EGFR;SDHD!E>!CDKN2A;STK11!E>!TP531 ! Li!Fraumeni!and!Related!Syndromes LUSC 0.03 CDKN2A,TP53 5.79EE15 TP53!E>!Fluoropyrimidine!Pathway,!Pharmacodynamics(TYMS,TP53);TP53!E>!chaperones!modulate!interferon!signaling!pathway(TP53,RB1);TP53!E>!apoptotic!signaling!in!response!to!dna!damage(BID,TP53);TP53!E>!Direct!p53!effectors(TP53,BID,EGFR,RB1,PTEN,CREBBP);TP53,CDKN2A,CHEK2!E>!p53!pathway(CDKN2A,TP53,CREBBP);CDKN2A!E>!CEMYC!pathway(CDKN2A,FBXW7);TP53!E>!estrogen!responsive!protein!efp!controls!cell!cycle!and!breast!tumors!growth(TP53,CDK6);CDKN2A,TP53!E>!Hypoxic!and!oxygen!homeostasis!regulation!of!HIFE1Ealpha(CDKN2A,TP53);TP53!E>!telomeres!telomerase!cellular!aging!and!immortality(TP53,RB1);TP53!E>!btg!family!proteins!and!cell!cycle!regulation(TP53,RB1);TP53!E>!Transcriptional!!activation!of!!cell!cycle!inhibitor!p21(TP53);TP53!E>!p53!signaling!pathway(TP53,RB1);TP53!E>!regulation!of!transcriptional!activity!by!pml(TP53,CREBBP,RB1);TP53!E>!rb!tumor!suppressor/checkpoint!signaling!in!response!to!dna!damage(TP53,RB1);CDKN2A!E>!E2F!transcription!factor!network(TYMS,CDKN2A,CREBBP,RB1);TP53!E>!Glucocorticoid!receptor0.651962919 0.02405 CDKN2A!E>!PTEN;CHEK2!E>!CDK6,TP53,PTEN;TP53!E>!CREBBP1 ! Lipoprotein!Deficiencies LUSC 1 0.885354678 0.097427692 APOB,LCAT,APOA1!E>!ENOSF1(1.81eE02);MTTP,APOB,LCAT,SAR1B,APOA1!E>!PROS1(1.81eE02);!E>!AKR1C2(2.16eE02);APOB,LCAT,APOA1!E>!SLC6A12(2.16eE02);MTTP,APOB,LCAT,APOA1!E>!SELO(1.87eE02)1 1 ! Retinitis!Pigmentosa LUSC 1 EYS 1 5.69EE14 KLHL7,SPATA7,CRB1,MERTK,PDE6B,CERKL,PRCD,FAM161A!E>!CLCN4(4.22eE02);TTC8,CRX,FSCN2,RDH12,PRPH2,RHO,CNGB1,CRB1,NRL,PDE6B,PDE6G,CERKL,GUCA1B,TULP1,NR2E3,ROM1,USH2A,RP1,PRCD,IMPG2,C2orf71,SAG,RBP3,FAM161A!E>!EYS(3.56eE16);TTC8!E>!COLEC12(1.19eE03);TTC8,CRX,FSCN2,PRPH2,CNGB1,KLHL7,SPATA7,CRB1,MERTK,PDE6B,CERKL,TULP1,ROM1,PRCD,IMPG2,C2orf71,RBP3,FAM161A!E>!KCNIP4(1.35eE03);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,CRB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,PROM1,CNGA1,RP1,PRCD,C2orf71,RPE65,SAG,RBP3,FAM161A,ABCA4!E>!CLUL1(2.31eE14);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,CRB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,CNGA1,RP1,PRCD,C2orf71,RPE65,SAG,RBP3,FAM161A,ABCA4!E>!SLC6A13(1.01eE12);TTC8,CRX,LRAT,FSCN2,RDH12,PRPH2,CNGB1,SPATA7,CRB1,MERTK,CERKL,TULP1,ROM1,PROM1,CNGA1,PRCD,IMPG2,C2orf71,RBP3,FAM161A!E>!UNC13B(4.65eE06);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,RP1,PRCD,RPE65,SAG,RBP3,ABCA4!E>!KCNJ13(4.97eE12)1 1 ! Haemophilia LUSC 1 0.00999936 F8,F9!E>!intrinsic!prothrombin!activation!pathway(COL4A5,PROS1)1 1 1 ! Chronic!Granulomatous!Disease PRAD 1 1 0.078898011 NCF2,NCF4,CYBA!E>!GPS2(1.49eE02);NCF2,NCF4,CYBB,CYBA!E>!SPOPL(1.38eE02);NCF2,NCF4,CYBB,CYBA!E>!COTL1(3.89eE02);!E>!CRISPLD2(4.13eE02);NCF2,NCF4,CYBB,CYBA!E>!TMUB2(1.62eE02);NCF2!E>!MFI2(2.93eE02);CYBA!E>!ARRDC1(2.54eE02);NCF2,NCF4,CYBB,CYBA!E>!HELZ2(1.23eE02);NCF2,NCF4,CYBB,CYBA!E>!PTEN(1.32eE02);NCF2,NCF4,CYBB!E>!PAK2(1.35eE02);NCF2,NCF4,CYBB,CYBA!E>!NFATC1(1.34eE02);NCF2,NCF4,CYBB,CYBA!E>!CRTC2(1.42eE02);NCF2,NCF4,CYBB,CYBA!E>!ADRBK1(1.57eE02);NCF2,NCF4,CYBA!E>!ZGPAT(1.40eE02);NCF2,NCF4!E>!EGR3(2.34eE02);NCF2,NCF4,CYBA!E>!GPR160(1.26eE02);NCF2,CYBB,CYBA!E>!HNMT(1.21eE02);NCF2,NCF4,CYBB!E>!SENP5(3.88eE02);NCF2,NCF4,CYBB,CYBA!E>!DNAJC5(1.35eE02);NCF2,NCF4,CYBB,CYBA!E>!ANKRD13D(1.27eE02);NCF2,NCF4,CYBB,CYBA!E>!CTDP1(1.30eE02);NCF4,CYBA!E>!CDKN1B(3.93eE02);CYBA!E>!POLD4(2.93eE02);CYBA!E>!RPS27(2.87eE02);NCF2,NCF4,CYBB,CYBA!E>!TPD52L2(1.12eE02);NCF2,NCF4,CYBA!E>!GMEB2(1.40eE02);NCF2,NCF4,CYBB,CYBA!E>!PQLC1(1.16eE02);NCF2,NCF4,CYBB,CYBA!E>!RGS19(2.46eE02);NCF2,NCF4,CYBB,CYBA!E>!TOR4A(1.38eE02);NCF2,NCF4,CYBB!E>!R1 1 ! Glycogenosis PRAD 1 1 0.078728665 PGAM2,PHKB,PHKA2,AGL!E>!SLC25A30(1.28eE02)1 1 ! Congenital!Ichthyosis PRAD 1 1 0.02654965 ALOX12B,SPINK5,CSTA,TGM1!E>!C9orf169(2.49eE03);ALOX12B,ALOXE3,CSTA,TGM1!E>!PARD6G(3.82eE02);ALOXE3,CSTA,ABCA12,TGM1!E>!ARRDC1(2.44eE02);NIPAL4,KRT2,LIPN,ABCA12!E>!NRARP(4.60eE03);SPINK5,CSTA,NIPAL4,ABCA12,TGM1!E>!PTK6(4.60eE03)1 1 ! Disorders!of!Phosphorous!Metabolism PRAD 0.69 SLC34A3 0.000406927 SLC34A3!E>!Type!II!Na+/Pi!cotransporters(SLC34A3) 1 1 1 ! Spinocerebellar!Ataxia PRAD 1 1 0.027197561 APTX,ATXN7,TDP1,SYNE1,ATM,TTBK2,TBP,ITPR1,SETX,SYT14,C10orf2,PPP2R2B!E>!CASP8AP2(3.80eE03);JPH3,ATXN2,PDYN,SYNE1,ATM,SPTBN2,TTBK2,KCNC3,PRNP,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B,ATXN1!E>!CREBL2(1.97eE02);JPH3,APTX,CACNA1A,ATXN2,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,PRKCG,FGF14,SYT14,PPP2R2B!E>!DLG1(5.25eE03);JPH3,TDP1,KCNC3,FGF14!E>!SAMD10(1.32eE02);JPH3,CACNA1A,ATXN2,PDYN,SPTBN2,TTBK2,KCNC3,PRKCG,FGF14,SYT14,PPP2R2B!E>!NSMF(9.90eE03);JPH3,ZNF592,CACNA1A,ATXN7,ATXN2,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,FGF14,SETX,PPP2R2B,ATXN1!E>!PHC3(2.61eE03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B!E>!GRIN1(3.68eE03);JPH3,CACNA1A,ATXN2,SYNE1,SPTBN2,TTBK2,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B!E>!MYT1(2.84eE03);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,ATXN10,PRKCG,FGF14,SYT14,PPP2R2B!E>!ZNF285(1.11eE02);JPH3,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN10,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B!E>!STMN3(2.84eE03);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ATXN1 1 ! Specified!Hamartoses PRAD 0.66 PTEN 0.062379972 STK11!E>!Metformin!Pathway,!Pharmacodynamic(SLC2A4,CRTC2);PTEN!E>!RhoA!signaling!pathway(CDKN1B,PTEN);PTEN!E>!pten!dependent!cell!cycle!arrest!and!apoptosis(CDKN1B,PTEN);PTEN!E>!Negative!regulation!of!the!PI3K/AKT!network(PTEN)0.57662662 1 1 ! Retinitis!Pigmentosa PRAD 1 1 6.10EE15 CRX,EYS,SPATA7,CRB1,MERTK,PDE6B,CERKL,PRCD,IMPG2,C2orf71,RBP3!E>!CREBL2(4.56eE02);CRX,FSCN2,PRPH2,CNGB1,KLHL7,EYS,SPATA7,CRB1,PDE6B,CERKL,TULP1,PRCD,IMPG2,C2orf71,RBP3,FAM161A!E>!MYT1(7.66eE06);SNRNP200,CRB1,PRPF31,CERKL,IDH3B,PRPF8,FAM161A!E>!UCKL1(1.68eE03);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,EYS,CRB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,PROM1,CNGA1,RP1,PRCD,C2orf71,RPE65,SAG,RBP3,FAM161A,ABCA4!E>!KCNG4(2.10eE17);CRX,LRAT,FSCN2,RDH12,PRPH2,RHO,CNGB1,EYS,CRB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,BEST1,PROM1,CNGA1,RP1,PRCD,C2orf71,RPE65,SAG,RBP3,FAM161A,ABCA4!E>!SAMD7(1.45eE17);RPGR,CA4,PRPF3,BEST1,RP2,TOPORS,SEMA4A!E>!GPR160(1.80eE02);SNRNP200,EYS!E>!THSD7B(1.89eE05);!E>!WFDC1(1.49eE05);!E>!NXPH2(5.64eE03);TTC8,RDH12,SPATA7,MERTK,CNGA1,FAM161A!E>!ZFHX3(1.68eE03);CRX,FSCN2,RDH12,SNRNP200,PRPH2,CNGB1,KLHL7,EYS,SPATA7,CRB1,PDE6B,CERKL,PRPF3,TULP1,ROM1,PRCD,IMPG2,C2orf71,TOPORS,RBP3,FAM161A!E>!PCMTD2(7.66eE06);ZNF513,KLHL7,SPATA7,CRB1,PDE6B,CERKL,FAM161A!E>!TMEM145(2.0.34981818 0.26722222 ! Long!QT!Syndrome READ 0.69 CACNA1C 3.14EE21 CACNA1C!E>!Nicotine!Pathway!(Chromaffin!Cell),!Pharmacodynamics(CACNA1C);CACNA1C!E>!Sympathetic!Nerve!Pathway!(PreE!and!PostE!Ganglionic!Junction)(CACNA1C,TH);CACNA1C!E>!AntiEdiabetic!Drug!Potassium!Channel!Inhibitors!Pathway,!Pharmacodynamics(PDX1,CACNA1C,INS)1 1 1 ! Chronic!Granulomatous!Disease READ 1 1 0.07714767 !E>!CRLF2(3.23eE02);NCF2,NCF4,CYBB,CYBA!E>!PRPF3(2.17eE02);NCF2,NCF4,CYBB,CYBA!E>!CACNA2D4(1.29eE02);NCF2!E>!KRAS(4.79eE02);NCF4,CYBA!E>!TRAPPC2(2.54eE02);NCF2,NCF4,CYBB,CYBA!E>!PLAGL2(1.33eE02);NCF2,NCF4,CYBB,CYBA!E>!CSF2RA(1.45eE02);NCF2,CYBB,CYBA!E>!TCF7L2(1.37eE02);NCF2,NCF4,CYBA!E>!TCEANC(2.36eE02);NCF2,NCF4,CYBB!E>!CTNNBL1(1.42eE02);NCF2,NCF4!E>!RNF40(3.72eE02);NCF2,NCF4,CYBB,CYBA!E>!IRF2(1.53eE02);NCF2,NCF4,CYBB,CYBA!E>!CR1(1.22eE02);NCF2,NCF4,CYBB,CYBA!E>!FBRS(1.23eE02);NCF2,CYBB,CYBA!E>!HS3ST3B1(1.16eE02);NCF4!E>!C17orf103(2.44eE02);NCF2,NCF4,CYBB,CYBA!E>!IL3RA(1.69eE02);!E>!ADK(3.11eE02);NCF2,CYBB!E>!RAB9A(1.57eE02);NCF2,CYBB!E>!EMP1(4.04eE02);NCF2!E>!SRCAP(3.53eE02);NCF2,NCF4,CYBB,CYBA!E>!RAB39A(1.60eE02);NCF2,NCF4,CYBB,CYBA!E>!MAP2K3(2.17eE02);NCF2,NCF4,CYBB,CYBA!E>!MOSPD2(1.16eE02);NCF2,NCF4,CYBB,CYBA!E>!MCL1(1.19eE02);NCF2!E>!P2RY8(1.27eE02)1 1 ! Glycogenosis READ 0.86 PHKG2 6.77EE12 AGL,PYGM,PHKA1,PHKB,PHKA2,PHKG2!E>!Glycogen!breakdown!(glycogenolysis)(PHKG2,GYG2)0.372442654 1 0.36782353 ! Inherited!Anomalies!of!the!Skin READ 1 0.004631074 TERT!E>!HIFE1Ealpha!transcription!factor!network(SMAD4,MCL1,SMAD3);TERT!E>!telomeres!telomerase!cellular!aging!and!immortality(TP53,KRAS);TERT!E>!Validated!targets!of!CEMYC!transcriptional!activation(SMAD4,TP53,SMAD3)0.10458489 1 1 ! Spinocerebellar!Ataxia READ 1 ATXN10 0.008075087 CACNA1A!E>!Sympathetic!Nerve!Pathway!(PreE!and!PostE!Ganglionic!Junction)(CACNA1C,TH);PRNP!E>!Glypican!1!network(HCK,FGFR1);CACNA1A!E>!AntiEdiabetic!Drug!Potassium!Channel!Inhibitors!Pathway,!Pharmacodynamics(PDX1,CACNA1C,INS);TBP!E>!Validated!targets!of!CEMYC!transcriptional!repression(ERBB2,SMAD4,SMAD3)0.013136049 ATXN7,TDP1,SYNE1,ATM,TTBK2,TBP,ITPR1,PPP2R2B!E>!FHIT(3.07eE03);!E>!PPP2R3B(4.28eE02);JPH3,ATXN2,TTBK2,TBP,PPP2R2B!E>!CDRT4(4.76eE02);JPH3,APTX,CACNA1A,ATXN2,PDYN,SYNE1,SPTBN2,TTBK2,KCNC3,PRNP,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B!E>!GPM6B(3.44eE03);ZNF592,POLG,ATXN7,TDP1,ATM,TBP,ITPR1,SETX,ATXN1!E>!PRPF3(8.48eE03);JPH3,ZNF592,CACNA1A,ATXN2,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B,ATXN1!E>!CCSER1(2.41eE03);JPH3,ZNF592,CACNA1A,ATXN7,ATXN2,PDYN,TDP1,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,SETX,SYT14,PPP2R2B,ATXN1!E>!WHSC1L1(1.11eE03);JPH3,ZNF592,CACNA1A,ATXN2,PDYN,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,PRKCG,FGF14,AFG3L2,SYT14,PPP2R2B!E>!PARK2(2.44eE03);JPH3,SPTBN2,TTBK2,KCNC3,ITPR1,PRKCG,FGF14,PPP2R2B!E>!ENSA(4.21eE02);JPH3,ZNF592,CACNA1A,ATXN2,PDYN,SYNE1,ATM,SPTBN2,TTBK2,KCNC3,ITPR1,PRKCG,FGF14,SYT14,PPP2R2B,ATXN1!E>!ZNF785(1.51eE03);JPH3,ZNF592,CACNA1A,ATXN7,SYNE1,ATM,SPTBN2,TTBK2,TBP,KCNC3,ITPR1,FGF14,SETX,SYT14,PPP2R2B,ATXN1!E>!CUL5(1.42eE03);POLG,ATXN7,TDP1,SYNE1,ATM,TBP,ITPR1,1 1 ! Severe!Combined!Immunodeficiency READ 1 1 0.017157155 IL2RG,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,DCLRE1C!E>!PRPF3(1.29eE02);ZAP70,IL2RG,CIITA,JAK3,RFXAP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!OFD1(1.32eE03);RFXAP,DCLRE1C!E>!FNTA(4.56eE02);PNP!E>!CACNA2D4(4.19eE02);ZAP70,IL2RG,JAK3,RFXAP,IL7R,CD3D,DCLRE1C!E>!CSTF2T(1.10eE02);!E>!DDX47(3.75eE02);IL2RG,JAK3,RFXANK,RFX5,PTPRC!E>!PRR14(6.34eE03);IL2RG,JAK3,ADA,PNP,RFX5,PTPRC,DCLRE1C!E>!PLAGL2(1.07eE02);ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!TCEANC(2.91eE03);ZAP70,IL2RG,JAK3,RFXAP,RFX5,IL7R,CD3D,DCLRE1C!E>!ASXL1(1.37eE03);PNP!E>!CTNNBL1(3.18eE02);IL2RG,JAK3,ADA,PNP,PTPRC!E>!HCK(4.98eE02);ZAP70,IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,IL7R,DCLRE1C!E>!IRF2(9.30eE03);IL2RG,JAK3,ADA,PNP,PTPRC!E>!HS3ST3B1(1.89eE02);JAK3,PTPRC!E>!C17orf103(4.06eE02);RFX5!E>!GPRC5D(1.63eE02);CIITA!E>!FLT3(1.41eE03);RFXAP,AK2,DCLRE1C!E>!FANCB(2.62eE02);CIITA,RFX5,DCLRE1C!E>!ADK(1.46eE02);PNP,PTPRC!E>!P2RY8(2.19eE03);RFXAP!E>!ANK1(4.68eE02)1 1 ! Lipoprotein!Deficiencies READ 1 1 0.075339054 MTTP,APOB,LCAT,SAR1B,APOA1!E>!KLC4(3.60eE02);APOB,LCAT,APOA1!E>!FCN3(2.07eE02);MTTP!E>!INSEIGF2(2.31eE02);APOB,ABCA1!E>!HS3ST3B1(2.70eE02);!E>!ADK(3.79eE02);MTTP,APOB,LCAT,APOA1!E>!ARSD(1.18eE02);MTTP,APOB,LCAT,SAR1B,APOA1!E>!ARSE(2.07eE02)1 1 ! Disorders!of!Urea!Cycle!Metabolism READ 1 1 0.088390959 ASS1,NAGS,ARG1,ASL,CPS1!E>!KLC4(4.33eE02);NAGS,ARG1,ASL,CPS1!E>!FCN3(2.80eE02);ASS1,NAGS,ARG1,ASL,CPS1!E>!ARSD(1.57eE02);ASS1,NAGS,ARG1,ASL,CPS1!E>!ARSE(2.80eE02)1 1 ! Retinitis!Pigmentosa READ 1 PRPF3 1 4.04EE13 !E>!IMMP2L(2.15eE05);!E>!DHRSX(1.34eE03);EYS,CERKL!E>!NKX6E3(1.42eE02);RPGR,CRX,FSCN2,RDH12,PRPH2,CNGB1,EYS,CRB1,TULP1,ROM1,RP2,PRCD,IMPG2,C2orf71,RBP3,SEMA4A!E>!CACNA2D4(1.12eE08);FSCN2,PRPH2,RHO,CNGB1,PDE6A,PDE6G,GUCA1B,RLBP1,RGR,NR2E3,ROM1,CNGA1,RP1,SAG,ABCA4!E>!EGFL6(4.38eE09);CRX,FSCN2,RDH12,PRPH2,CNGB1,EYS,CRB1,CERKL,TULP1,ROM1,PRCD,IMPG2,C2orf71,RBP3!E>!ASMT(1.12eE08);TTC8,CRX,FSCN2,RDH12,PRPH2,CNGB1,KLHL7,EYS,SPATA7,CRB1,MERTK,PDE6B,CERKL,TULP1,ROM1,PRCD,IMPG2,C2orf71,RBP3,FAM161A!E>!KIAA1467(6.02eE04);!E>!ZBED1(3.70eE03);TTC8,CRX,FSCN2,PRPH2,CNGB1,KLHL7,EYS,SPATA7,CRB1,TULP1,PRCD,IMPG2,C2orf71,RBP3,FAM161A!E>!GSG1(6.99eE07);CRX,FSCN2,RDH12,PRPH2,RHO,CNGB1,NRL,PDE6B,PDE6G,GUCA1B,TULP1,NR2E3,ROM1,PROM1,CNGA1,RP1,PRCD,C2orf71,SAG,RBP3,FAM161A,ABCA4!E>!KERA(5.05eE15);FAM161A!E>!SHOX(9.94eE11)1 0.40083333 ! DopaEResponsive!Dystonia READ 0.62 TH 2.34EE32 TH!E>!AlphaEsynuclein!signaling(HCK,TH,PARK2);TH!E>!Sympathetic!Nerve!Pathway!(PreE!and!PostE!Ganglionic!Junction)(CACNA1C,TH)1 1 1 ! Congenital!Ectodermal!Dysplasia SKCM 1 0.28548436 0.000106457 !E>!FAM58A(9.62eE03);LAMB3,ITGB4,PLEC,KRT5,LAMC2!E>!ANAPC15(2.97eE03);!E>!EIF3D(1.58eE03);ITGA6,LAMB3,ITGB4,GJB6,COL17A1,PLEC,COL7A1,KRT5,KRT14,LAMC2,LAMA3!E>!PTK6(2.00eE06);LAMB3,ITGB4,KRT5!E>!MYC(2.50eE04);!E>!CDKN2A(2.84eE02);PLEC!E>!PHGDH(3.27eE02);PLEC!E>!RPL13(4.57eE03);!E>!CCND1(2.48eE02);LAMB3,ITGB4,PLEC,KRT5!E>!PPDPF(2.47eE03);!E>!LSM12(2.57eE02);!E>!SLC25A39(1.42eE02);!E>!MRP63(4.21eE02);!E>!SRMS(2.88eE02);ITGA6,LAMB3,ITGB4,COL17A1,KRT5,KRT14,LAMC2,LAMA3!E>!TDRP(5.01eE04);PLEC!E>!GRN(1.79eE02);!E>!TP53(1.56eE02);!E>!TSPAN31(1.19eE02);!E>!HDAC3(3.26eE03);ITGB4,PLEC,KRT5!E>!ACD(5.04eE03);!E>!KRT78(1.13eE02);GJB6!E>!TCHHL1(9.83eE03);PLEC!E>!DEF8(2.72eE02);ITGA6,COL17A1,COL7A1,KRT14!E>!DSG1(1.08eE05);LAMB3,ITGB4,PLEC,KRT5,KRT14,LAMC2!E>!TNFRSF6B(6.12eE04);!E>!RPTN(3.38eE04);PLEC!E>!TPD52L2(4.98eE02);LAMB3,ITGB4,PLEC,COL7A1,KRT5!E>!CHMP1A(1.44eE02);!E>!DYNAP(9.82eE03);ITGB4,PLEC,COL7A1!E>!SLC2A4RG(6.70eE03);!E>!FOLR3(2.43eE03)1 1 ! Chronic!Granulomatous!Disease SKCM 1 1 0.078763609 NCF2,NCF4,CYBB,CYBA!E>!ZFX(1.54eE02);NCF2,NCF4,CYBB,CYBA!E>!HELZ2(1.79eE02);NCF2,NCF4!E>!KIAA1257(2.33eE02);NCF2!E>!STK19(3.34eE02);NCF2,NCF4,CYBB,CYBA!E>!B2M(2.08eE02);NCF2,NCF4,CYBB!E>!VPS9D1(1.84eE02);NCF2,NCF4,CYBB,CYBA!E>!DDX3X(1.31eE02);NCF2,NCF4,CYBB,CYBA!E>!TMUB2(1.76eE02);NCF2,NCF4,CYBB,CYBA!E>!CLEC2B(2.70eE02);NCF2,NCF4,CYBB,CYBA!E>!ADAM8(1.31eE02);NCF2,NCF4,CYBB,CYBA!E>!PTEN(1.49eE02);NCF2,NCF4,CYBB,CYBA!E>!ZNF276(1.43eE02);NCF4!E>!SLC25A39(1.56eE02);NCF2,NCF4,CYBB,CYBA!E>!GNAI2(1.51eE02);NCF2,NCF4,CYBB,CYBA!E>!PPP6C(1.49eE02);NCF2,NCF4,CYBB,CYBA!E>!RPGRIP1(1.45eE02);NCF4,CYBB!E>!POM121(3.52eE02);NCF2,NCF4,CYBA!E>!ZGPAT(1.56eE02);!E>!SERPINB10(1.35eE02);NCF2,NCF4,CYBB,CYBA!E>!GRN(3.32eE02);NCF2,NCF4,CYBB,CYBA!E>!DNAJC5(1.52eE02);NCF2,NCF4,CYBB,CYBA!E>!RBM22(1.44eE02);NCF4!E>!ITGA2B(3.33eE02);CYBB,CYBA!E>!OXA1L(1.33eE02);NCF2,NCF4,CYBB!E>!MPP7(1.54eE02);!E>!SLC4A1(3.39eE02);NCF2,NCF4,CYBB,CYBA!E>!CLEC5A(1.39eE02);NCF2,NCF4,CYBB,CYBA!E>!GPR141(1.49eE02);NCF4,CYBB,CYBA!E>!ITGA4(1.49eE02);NCF2,NCF4,CYB1 1 ! Congenital!Ichthyosis SKCM 1 1 0.005136028 SPINK5,CSTA,NIPAL4,ABCA12,TGM1!E>!PTK6(3.02eE03);SPINK5,CSTA,NIPAL4,KRT2,LIPN,ABHD5!E>!MPP7(1.32eE02);ALOX12B,SPINK5,CSTA,TGM1!E>!KRT78(5.11eE03);ALOX12B,ABCA12!E>!TCHHL1(3.36eE04);ALOX12B,SPINK5,CSTA,ABCA12,TGM1!E>!DSG1(2.89eE04);ALOX12B,SPINK5,CSTA,ABCA12,TGM1!E>!RPTN(2.89eE04);ALOX12B,SPINK5,CSTA,TGM1!E>!DYNAP(3.02eE03)1 1 ! Polycystic!Kidney,!Autosomal!Dominant SKCM 1 0.013304666 TSC2!E>!LKB1!signaling!events(RPTOR,MYC,TP53);TSC2!E>!Validated!targets!of!CEMYC!transcriptional!repression(CCND1,HDAC3,EP300,MYC)0.736679678 1 1 ! Inherited!Anomalies!of!the!Skin SKCM 1 TERT 5.99EE19 TERT!E>!erk1/erk2!mapk!signaling!pathway(MYC,TERT);TERT!E>!IL2!signaling!events!mediated!by!PI3K(MYC,TERT,RAC1);TERT!E>!overview!of!telomerase!protein!component!gene!htert!transcriptional!regulation(MYC,TERT,TP53);TERT!E>!telomeres!telomerase!cellular!aging!and!immortality(MYC,TERT,TP53);KRT1,TERT!E>!Regulation!of!nuclear!beta!catenin!signaling!and!target!gene!transcription(TERT,CCND1,CDKN2A,MITF,MYC,EP300);TERT!E>!role!of!nicotinic!acetylcholine!receptors!in!the!regulation!of!apoptosis(TERT,FASLG);TERT!E>!Validated!targets!of!CEMYC!transcriptional!activation(TERT,TP53,UBTF,CDK4,MYC,EP300);TINF2,TERT,DKC1!E>!Regulation!of!Telomerase(CCND1,MYC,TERT,SAP18,ACD)0.164716031 1 1 ! Combined!Heart!and!Skeletal!Defects SKCM 0.64 EP300 2.01EE19 EP300!E>!cell!cycle:!g2/m!checkpoint(EP300,TP53,MYT1);EP300!E>!Validated!transcriptional!targets!of!AP1!family!members!Fra1!and!Fra2(CCND1,EP300,CDKN2A);EP300,CREBBP!E>!acetylation!and!deacetylation!of!rela!in!nucleus(HDAC3,EP300);EP300!E>!Notch!signaling!pathway(CCND1,MYC,EP300,NOTCH2,DNER);EP300!E>!Validated!nuclear!estrogen!receptor!alpha!network(CCND1,KLRC3,EP300,MYC);EP300!E>!Regulation!of!nuclear!beta!catenin!signaling!and!target!gene!transcription(TERT,CCND1,CDKN2A,MITF,MYC,EP300);EP300!E>!Validated!targets!of!CEMYC!transcriptional!repression(CCND1,HDAC3,EP300,MYC);EP300,CREBBP!E>!E2F!transcription!factor!network(CCNE2,EP300,MYC,CDKN2A);EP300,CREBBP!E>!ilE7!signal!transduction(ITGA2B,EP300);EP300,CREBBP!E>!Validated!targets!of!CEMYC!transcriptional!activation(TERT,TP53,UBTF,CDK4,MYC,EP300);EP300!E>!hypoxia!and!p53!in!the!cardiovascular!system(EP300,TP53);EP300!E>!ATFE2!transcription!factor!network(CCND1,EP300,CDK4,NF1);EP300!E>!melanocyte!development!and!pigmentation!pathway(MITF,EP300);EP300,CREBBP!E>1 0.32066667 1 ! Specified!Anomalies!of!the!Musculoskeletal!SystemSKCM 1 MITF 0.097075171 MITF,SNAI2!E>!Regulation!of!nuclear!beta!catenin!signaling!and!target!gene!transcription(TERT,CCND1,CDKN2A,MITF,MYC,EP300);MITF!E>!IL6Emediated!signaling!events(MITF,MYC,RAC1);MITF!E>!melanocyte!development!and!pigmentation!pathway(MITF,EP300)1 1 1 ! Neurofibromatosis SKCM 0.61 NF1 4.53EE05 NF1!E>!Regulation!of!Ras!family!activation(NRAS,RASA2,NF1);NF1!E>!ATFE2!transcription!factor!network(CCND1,EP300,CDK4,NF1)0.719411434 0.222 NF2!E>!CDKN2A 1 ! Tuberous!Sclerosis SKCM 1 0.013304666 TSC2,TSC1!E>!LKB1!signaling!events(RPTOR,MYC,TP53);TSC2!E>!Validated!targets!of!CEMYC!transcriptional!repression(CCND1,HDAC3,EP300,MYC)0.594661126 1 1 ! Severe!Combined!Immunodeficiency SKCM 1 1 0.001386824 IL2RG,JAK3,ADA,PNP,PTPRC!E>!HELZ2(4.66eE02);ZAP70,IL2RG,JAK3,PTPRC,IL7R,CD3D,DCLRE1C,RAG2,RAG1!E>!THEMIS(1.76eE04);ZAP70,IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!B2M(6.45eE03);NHEJ1,RFX5,AK2!E>!EIF3D(4.81eE05);ZAP70,IL2RG,JAK3,ADA,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!KCNAB2(3.23eE03);ZAP70,RFXANK,IL7R,CD3D!E>!RPL13(2.74eE03);AK2!E>!TERT(1.68eE02);ZAP70,IL2RG,JAK3,PTPRC,IL7R,CD3D,DCLRE1C!E>!TC2N(5.55eE03);ZAP70,IL2RG,JAK3,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!CLEC2B(3.84eE03);ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!CLEC2D(1.81eE04);IL2RG,JAK3,PNP,RFX5,PTPRC!E>!ADAM8(3.98eE02);ZAP70,IL2RG,CIITA,JAK3,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!ZNF276(3.22eE03);ZAP70,IL2RG,JAK3,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!PCED1B(1.76eE03);JAK3!E>!SLC25A39(7.97eE04);IL2RG,CIITA,PNP,RFX5,PTPRC,DCLRE1C!E>!LY86(1.75eE02);ZAP70,IL2RG,JAK3,PNP,RFX5,PTPRC,DCLRE1C!E>!PPP6C(5.65eE03);ZAP70,IL2RG,JAK3,RFXAP,PNP,RFX5,PTPRC,IL7R,CD3D,DCLRE1C!E>!POM121(8.56eE04);ZAP70,IL2RG,JAK3,ADA,PTPRC,IL7R,CD3D!E>!ZGPAT(2.74eE03);PNP,AK2!E>!1 1 ! Specified!Hamartoses SKCM 0.72 PTEN 0.061006438 STK11!E>!LKB1!signaling!events(RPTOR,MYC,TP53);VHL!E>!Hypoxic!and!oxygen!homeostasis!regulation!of!HIFE1Ealpha(CDKN2A,TP53);STK11!E>!Regulation!of!AMPK!activity!via!LKB1(RPTOR);PTEN!E>!pten!dependent!cell!cycle!arrest!and!apoptosis(PTEN,FASLG)0.561087039 0.13528125 PTEN!E>!TP53;SDHB!E>!CDKN2A;STK11!E>!NRAS,OGDHL,TP530.2405 PTEN!E>!TCEB3C;SDHB!E>!TP53;STK11!E>!CCNE2;VHL!E>!CHGB! Li!Fraumeni!and!Related!Syndromes SKCM 0.05 CDKN2A,TP53 2.63EE31 TP53!E>!LKB1!signaling!events(RPTOR,MYC,TP53);TP53,CHEK2!E>!cell!cycle:!g2/m!checkpoint(EP300,TP53,MYT1);CDKN2A!E>!Validated!transcriptional!targets!of!AP1!family!members!Fra1!and!Fra2(CCND1,EP300,CDKN2A);TP53!E>!estrogen!responsive!protein!efp!controls!cell!cycle!and!breast!tumors!growth(CDK4,TP53);TP53!E>!overview!of!telomerase!protein!component!gene!htert!transcriptional!regulation(MYC,TERT,TP53);TP53,CHEK2!E>!role!of!brca1!brca2!and!atr!in!cancer!susceptibility(TP53,FANCA);CDKN2A,TP53!E>!Hypoxic!and!oxygen!homeostasis!regulation!of!HIFE1Ealpha(CDKN2A,TP53);TP53!E>!telomeres!telomerase!cellular!aging!and!immortality(MYC,TERT,TP53);CDKN2A!E>!Regulation!of!nuclear!beta!catenin!signaling!and!target!gene!transcription(TERT,CCND1,CDKN2A,MITF,MYC,EP300);TP53!E>!btg!family!proteins!and!cell!cycle!regulation(CCND1,TP53);TP53!E>!Transcriptional!!activation!of!!cell!cycle!inhibitor!p21(TP53);TP53!E>!p53!signaling!pathway(CCND1,CDK4,TP53);TP53!E>!rb!tumor!suppressor/checkpoint!signaling!in!response!to!dna!damage(CDK40.662892321 0.15561765 CDKN2A!E>!PTEN;CHEK2!E>!CDK4,TP53,PTEN;TP53!E>!EP3001 ! Lipoprotein!Deficiencies SKCM 1 1 0.058815075 APOB,LCAT,APOA1!E>!SPTLC3(3.68eE02);APOB,SAR1B!E>!IDH1(2.34eE02);MTTP,APOB,LCAT,SAR1B,APOA1!E>!UGT2B15(6.62eE03);MTTP,APOB,LCAT,SAR1B,APOA1,ABCA1!E>!C2(1.13eE02);APOB,APOA1!E>!TIMD4(3.68eE02);APOB,APOA1!E>!STAB2(1.22eE02);MTTP,APOB,LCAT,SAR1B,APOA1!E>!APCS(1.13eE02)1 1 ! Retinitis!Pigmentosa SKCM 1 EYS,CRB1 1 2.00EE13 CERKL,IDH3B!E>!ASB16(5.43eE03);CRX,FSCN2,KLHL7,SPATA7,MERTK,PDE6B,CERKL,ROM1,PRCD,C2orf71,RBP3,FAM161A!E>!SV2B(3.21eE03);CRX,FSCN2,PRPH2,CNGB1,KLHL7,SPATA7,PDE6B,CERKL,TULP1,PRCD,IMPG2,C2orf71,RBP3,FAM161A!E>!MYT1(4.84eE05);TTC8,CRX,FSCN2,RDH12,PRPH2,RHO,CNGB1,KLHL7,SPATA7,NRL,MERTK,PDE6B,PDE6G,CERKL,GUCA1B,TULP1,NR2E3,ROM1,PRCD,IMPG2,C2orf71,SAG,RBP3,FAM161A,ABCA4!E>!CRB1(2.61eE05);CRX,FSCN2,RDH12,PRPH2,CNGB1,CERKL,TULP1,ROM1,PRCD,IMPG2,C2orf71,RBP3!E>!LRTM1(5.49eE09);TTC8,CRX,FSCN2,RDH12,PRPH2,CNGB1,SPATA7,PDE6B,TULP1,ROM1,PRCD,RBP3!E>!FMN1(2.31eE04);CRX,FSCN2,RDH12,PRPH2,RHO,CNGB1,NRL,PDE6B,PDE6A,PDE6G,GUCA1B,RLBP1,TULP1,RGR,NR2E3,ROM1,PROM1,CNGA1,RP1,PRCD,C2orf71,SAG,RBP3,FAM161A,ABCA4!E>!PROL1(6.23eE14);KLHL7!E>!NUDT4(2.32eE02);KLHL7,IDH3B!E>!SCN5A(4.86eE02);LRAT,SNRNP200,PROM1!E>!UTF1(4.41eE02);ZNF513,KLHL7,SPATA7,MERTK,PDE6B,PRCD,FAM161A!E>!CHRNA4(1.59eE02);CRX,SNRNP200,CA4,CERKL,TULP1,PROM1,FAM161A!E>!KCNB2(9.21eE04);TTC8,RPGR,SPATA7,CERKL,FAM161A!E>!CHGB(4.86eE02);!E>!MITF(1.67eE04);!E>!SPTBN5(4.63eE0.21645 IDH3B!E>!IDH1,EIF2B1;IMPDH1!E>!SKIV2L;PROM1!E>!DDX3X;PRPF3!E>!RPL13,CRB1,PRPF6,PRPF6,DDX3X;PRPF31!E>!PRPF6,RPGRIP1,MYC;PRPF8!E>!PRPF60.22446667 CA4!E>!SLC4A1,EP300;CRX!E>!RBFOX1;FAM161A!E>!PARK2,RUNDC3A,MYC;IDH3B!E>!TP53,DDX3X,HDAC3,PRPF6,FANCA;IMPDH1!E>!PRPF6,DDX3X;NR2E3!E>!ITGA4;PRPF3!E>!HDAC5;PRPF31!E>!MYC;PRPF8!E>!PRPF6,PHGDH,RPGRIP1,HDAC5;ROM1!E>!ITGA4;RPGR!E>!PARK2;SNRNP200!E>!MYC;TOPORS!E>!PRPF6,UPF3A! Hereditary!Hemorrhagic!Telangiectasia SKCM 1 1.86EE06 SMAD4!E>!LKB1!signaling!events(RPTOR,MYC,TP53);SMAD4!E>!Validated!nuclear!estrogen!receptor!alpha!network(CCND1,KLRC3,EP300,MYC);SMAD4!E>!Validated!targets!of!CEMYC!transcriptional!repression(CCND1,HDAC3,EP300,MYC);SMAD4!E>!Validated!targets!of!CEMYC!transcriptional!activation(TERT,TP53,UBTF,CDK4,MYC,EP300)1 1 1 ! Disorders!of!Aromatic!Amino!Acid!MetabolismSKCM 1 MC1R 1 0.072585714 BLOC1S6,AP3B1,BLOC1S3,HPS5,HPS1!E>!HELZ2(1.61eE02);BLOC1S6,AP3B1,BLOC1S3,HPS5,HPS6,HPS1!E>!TMUB2(3.76eE02);TYR,OCA2,TYRP1,SLC45A2,BLOC1S6,AP3B1,DTNBP1,BLOC1S3,HPS5,HPS1!E>!KCNAB2(9.33eE03);BLOC1S6,AP3B1,BLOC1S3,HPS6,HPS1!E>!GNAI2(2.33eE02);BLOC1S6,BLOC1S3,HPS5,HPS6,HPS1!E>!GRN(1.01eE02);!E>!OXA1L(2.24eE02);BLOC1S6,AP3B1,HPS5!E>!CLEC5A(3.63eE02);BLOC1S6,BLOC1S3,HPS5,HPS6,HPS1!E>!TPD52L2(3.55eE02);TAT,HPD,FAH!E>!C2(3.91eE02);HPD!E>!LIME1(4.55eE02);BLOC1S6,AP3B1,BLOC1S3,HPS5,HPS1!E>!RGS19(3.70eE02);TYR,OCA2,TYRP1,SLC45A2!E>!MITF(2.04eE02);BLOC1S3,HPS6,HPS1!E>!INPPL1(2.36eE02);TYR,OCA2,TYRP1,SLC45A2,BLOC1S3,HPS5,HPS1!E>!MAD1L1(1.01eE02);BLOC1S6,DTNBP1,BLOC1S3,HPS1!E>!TCF25(3.81eE02);BLOC1S6,AP3B1,DTNBP1,BLOC1S3,HPS5,HPS1!E>!SBNO2(2.33eE02);HPS1!E>!FOLR2(1.98eE02)1 1 ! Chronic!Granulomatous!Disease STAD 1 0.846838069 0.072585714 NCF2,NCF4,CYBB,CYBA!E>!B2M(2.04eE02);NCF2,NCF4,CYBB,CYBA!E>!DIAPH2(1.14eE02);NCF2,NCF4,CYBB,CYBA!E>!TRPS1(9.36eE03);NCF2!E>!KRAS(4.90eE02);CYBA!E>!CLECL1(3.26eE02);NCF2,NCF4,CYBB,CYBA!E>!CLEC2B(2.77eE02);NCF2,NCF4,CYBB,CYBA!E>!PTEN(1.14eE02);NCF2,NCF4,CYBB,CYBA!E>!SNX2(1.75eE02);NCF2,NCF4,CYBB,CYBA!E>!CLEC12A(1.60eE02);NCF4!E>!IL5RA(3.13eE02);NCF2,CYBB,CYBA!E>!RHOA(9.22eE03);NCF2,NCF4,CYBB,CYBA!E>!IRF2(9.69eE03);!E>!KLRF1(3.81eE02);!E>!CLEC1B(3.30eE02);NCF4,CYBA!E>!CD69(3.41eE02);NCF2,NCF4,CYBB,CYBA!E>!DYRK1A(2.15eE02);NCF2,CYBB,CYBA!E>!UAP1L1(1.35eE02);!E>!PLGRKT(3.41eE02);CYBB,CYBA!E>!DPP7(1.97eE02);NCF2,CYBB,CYBA!E>!CD44(9.47eE03);NCF2,NCF4,CYBB,CYBA!E>!CD274(1.06eE02)1 1 ! Disorders!of!Phosphorous!Metabolism STAD 0.67 SLC34A3 0.003812225 FGF23!E>!SyndecanE2Emediated!signaling!events(RHOA,FGFR2,FGF19);FGF23!E>!SyndecanE3Emediated!signaling!events(FGFR2,FGF19,EGFR);SLC34A3!E>!Type!II!Na+/Pi!cotransporters(SLC34A3);FGF23!E>!FGF!signaling!pathway(CDH1,PIK3CA,FGF19,FGFR2)1 0.11223333 FGF23!E>!FGF19 1 ! Combined!Heart!and!Skeletal!Defects STAD 1 0.002558855 CREBBP!E>!the!information!processing!pathway!at!the!ifn!beta!enhancer(IRF2,ARID1A);EP300!E>!Regulation!of!nuclear!beta!catenin!signaling!and!target!gene!transcription(CDH1,APC,AES,CDKN2A);EP300!E>!p73!transcription!factor!network(RNF43,CDK6,WWOX)0.650459398 0.04182609 CREBBP!E>!TP53,TP53,GATA40.26054167 ! Specified!Hamartoses STAD 0.63 PTEN 1.38EE07 PTEN!E>!skeletal!muscle!hypertrophy!is!regulated!via!aktEmtor!pathway(PIK3CA,PTEN);PTEN!E>!regulation!of!eifE4e!and!p70s6!kinase(PIK3CA,PTEN);PTEN!E>!mtor!signaling!pathway(PTEN,PIK3CA);VHL!E>!Hypoxic!and!oxygen!homeostasis!regulation!of!HIFE1Ealpha(CDKN2A,TP53);PTEN!E>!RhoA!signaling!pathway(PTEN,RHOA,MAP2K4);PTEN!E>!pten!dependent!cell!cycle!arrest!and!apoptosis(PTEN,PIK3CA);PTEN!E>!Negative!regulation!of!the!PI3K/AKT!network(PTEN)0.719559223 0 PTEN!E>!PIK3CA;STK11!E>!EGFR1 ! Li!Fraumeni!and!Related!Syndromes STAD 0.03 CDKN2A,TP53 1.68EE20 CDKN2A!E>!CEMYC!pathway(CDKN2A,FBXW7);TP53!E>!estrogen!responsive!protein!efp!controls!cell!cycle!and!breast!tumors!growth(TP53,CDK6);CDKN2A,TP53!E>!Hypoxic!and!oxygen!homeostasis!regulation!of!HIFE1Ealpha(CDKN2A,TP53);TP53!E>!telomeres!telomerase!cellular!aging!and!immortality(TP53,KRAS);CDKN2A!E>!Regulation!of!nuclear!beta!catenin!signaling!and!target!gene!transcription(CDH1,APC,AES,CDKN2A);TP53!E>!BARD1!signaling!events(CCNE1,TP53);TP53!E>!Transcriptional!!activation!of!!cell!cycle!inhibitor!p21(TP53);TP53,CHEK2!E>!PLK3!signaling!events(CCNE1,TP53);TP53!E>!p53!signaling!pathway(CCNE1,TP53);TP53!E>!p75(NTR)Emediated!signaling(PIK3CA,RHOA,TP53)0.693511719 0 CDKN2A!E>!PTEN;CHEK2!E>!CDK6,TP53,SMAD4;TP53!E>!PTEN1 ! Chronic!Granulomatous!Disease UCEC 1 1 0.078663977 NCF2,NCF4,CYBA!E>!GMEB2(1.50eE02);NCF2,NCF4,CYBA!E>!ZNF263(1.79eE02);NCF2,CYBA!E>!IRAK1(2.62eE02);NCF2,NCF4,CYBB,CYBA!E>!HELZ2(1.44eE02);NCF2!E>!KRAS(4.41eE02);NCF2,NCF4,CYBB,CYBA!E>!ADAMDEC1(4.66eE02);NCF2,NCF4,CYBB,CYBA!E>!PTEN(1.31eE02);NCF2,NCF4,CYBB,CYBA!E>!PLAGL2(1.18eE02);NCF2,NCF4,CYBB,CYBA!E>!NFATC1(1.48eE02);NCF2,NCF4,CYBB,CYBA!E>!VDR(1.25eE02);NCF2,NCF4,CYBB,CYBA!E>!DNM2(1.64eE02);!E>!HAUS8(4.54eE02);NCF2,NCF4,CYBA!E>!ZGPAT(1.41eE02);NCF4,CYBB,CYBA!E>!NEK8(1.32eE02);NCF2,NCF4,CYBB,CYBA!E>!NFE2L2(1.19eE02);NCF2,NCF4,CYBB,CYBA!E>!CREBBP(1.30eE02);CYBA!E>!TMEM80(2.60eE02);NCF2,NCF4,CYBB,CYBA!E>!CTDP1(1.26eE02);NCF2,NCF4,CYBA!E>!CHMP2A(1.31eE02);CYBA!E>!RPLP2(4.66eE02);NCF2,NCF4,CYBB,CYBA!E>!SAP30BP(1.29eE02);CYBA!E>!POLD4(2.83eE02);NCF2,NCF4,CYBB,CYBA!E>!TPD52L2(1.15eE02);NCF4,CYBB,CYBA!E>!ADAM28(1.19eE02);NCF2,NCF4,CYBB,CYBA!E>!PQLC1(1.18eE02);NCF2,CYBB!E>!CTNNB1(3.35eE02);NCF2,NCF4,CYBB,CYBA!E>!SGK1(1.18eE02);NCF2,NCF4,CYBB,CYBA!E>!MYO9B(1.42eE02);NCF2,CYBB,CYBA!E>!NEU4(1.31eE02);NCF2,NCF4,CYBB,CYBA1 1 ! DiamondEBlackfan!Anemia UCEC 1 1.14EE06 RPS26,RPS24,RPS10,RPS17,RPS19,RPL5,RPL35A,RPS7,RPL11!E>!Regulation!of!gene!expression!in!beta!cells(RPL14,RPLP2,RPS5,FOXA2,RPL22);RPL11!E>!Validated!targets!of!CEMYC!transcriptional!activation(TAF4B,TERT,TP53,MYC,CREBBP)0.00122045 RPL5,RPS7!E>!NRAS(1.39eE03);RPS19,RPL35A,RPS7!E>!MYC(2.25eE04);RPS19!E>!IRAK1(3.08eE02);RPS26,RPS19,RPS10,RPL35A,RPS7!E>!TRIM28(1.64eE02);RPS26,RPS7!E>!HAUS8(1.64eE03);!E>!CCNE1(6.83eE03);RPS26,RPS7!E>!FTSJ2(1.78eE02);RPS26,RPS19,RPS10,RPL35A,RPS7!E>!NUDT1(2.14eE03);RPS26,RPS24,RPL5,RPS19,RPS10,RPL11,RPL35A,RPS7!E>!RPS5(3.05eE05);RPS24,RPL5,RPS19,RPS10,RPL11,RPL35A!E>!RPLP2(3.96eE04);RPS19,RPS10!E>!POLD4(2.33eE02);RPS26,RPS19,RPS7!E>!TP53(1.72eE03);RPS26,RPS19,RPS7!E>!TACC3(8.31eE04);RPS26,RPS19,RPS7!E>!GEMIN4(2.74eE02);RPS26,RPS10,RPL35A!E>!RNMTL1(3.46eE04);RPS10,RPL35A!E>!ZNF497(1.11eE03);RPS26,RPS19,RPS10,RPL35A,RPS7!E>!TXNL4A(3.23eE02);RPS24,RPL5,RPS19,RPS10,RPL11,RPL35A,RPS7!E>!RPL22(3.92eE05);RPS10!E>!C9orf142(3.77eE03);!E>!TERT(1.23eE02);RPS19!E>!TRAF4(3.89eE02);RPS24,RPL5,RPS19,RPS10,RPL11,RPL35A,RPS7!E>!RPL14(2.97eE04);RPL11,RPL35A,RPS7!E>!RBMX(2.78eE02)0 RPL11!E>!FHIT;RPL35A!E>!MECOM;RPL5!E>!RPLP2;RPS10!E>!RPL14;RPS17!E>!RPS5;RPS19!E>!WWOX,FHIT;RPS24!E>!MECOM;RPS26!E>!RPLP2;RPS7!E>!RPL140.20041667 RPL11!E>!RPL14;RPL5!E>!TP53;RPS10!E>!ESR1;RPS17!E>!RPL22;RPS19!E>!RPS5;RPS24!E>!MYC,RPL14;RPS26!E>!ESR1;RPS7!E>!RPLP2! Inherited!Anomalies!of!the!Skin UCEC 1 TERT 3.29EE08 ATP2A2!E>!nfat!and!hypertrophy!of!the!heart!(NFATC1,CREBBP,PIK3R1,PIK3CA);TERT!E>!IL2!signaling!events!mediated!by!PI3K(PIK3CA,MYC,TERT,PIK3R1);TERT!E>!overview!of!telomerase!protein!component!gene!htert!transcriptional!regulation(MYC,TERT,TP53,MZF1,ESR1);TERT!E>!telomeres!telomerase!cellular!aging!and!immortality(MYC,TERT,TP53,RB1,KRAS);KRT1,TERT!E>!Regulation!of!nuclear!beta!catenin!signaling!and!target!gene!transcription(CTNNB1,CCND1,MYC,TERT);TERT!E>!role!of!nicotinic!acetylcholine!receptors!in!the!regulation!of!apoptosis(PIK3CA,TERT,PIK3R1);TERT!E>!Validated!targets!of!CEMYC!transcriptional!activation(TAF4B,TERT,TP53,MYC,CREBBP);TINF2,TERT,DKC1!E>!Regulation!of!Telomerase(CCND1,MYC,TERT,SIN3A,ESR1)0.183228108 1 1 ! Combined!Heart!and!Skeletal!Defects UCEC 0.63 CREBBP 2.76EE30 CREBBP!E>!nfat!and!hypertrophy!of!the!heart!(NFATC1,CREBBP,PIK3R1,PIK3CA);EP300,CREBBP!E>!IFNEgamma!pathway(PIK3CA,CREBBP,PIK3R1);EP300,CREBBP!E>!Direct!p53!effectors(VDR,TP53,PIDD,RB1,PTEN,CREBBP);EP300,CREBBP!E>!FOXA1!transcription!factor!network(FOXA2,CREBBP,NKX3E1,ESR1);EP300,CREBBP!E>!transcription!regulation!by!methyltransferase!of!carm1(CREBBP,PRKAR1B);EP300!E>!cell!cycle:!g2/m!checkpoint(TP53,MYT1);EP300,CREBBP!E>!carm1!and!regulation!of!the!estrogen!receptor(CREBBP,ESR1);CREBBP!E>!wnt!signaling!pathway(CTNNB1,CCND1,MYC,CREBBP);EP300!E>!Validated!nuclear!estrogen!receptor!alpha!network(CCND1,MYC,UBE2M,ESR1);EP300!E>!Regulation!of!nuclear!beta!catenin!signaling!and!target!gene!transcription(CTNNB1,CCND1,MYC,TERT);EP300,CREBBP!E>!mechanism!of!gene!regulation!by!peroxisome!proliferators!via!ppara(MYC,PRKAR1B,RB1,CREBBP);CREBBP!E>!regulation!of!transcriptional!activity!by!pml(TP53,CREBBP,RB1);EP300,CREBBP!E>!E2F!transcription!factor!network(MYC,CCNE1,TRIM28,CREBBP,RB1);EP300,CREBBP!E>!ilE7!signal!transduc1 0.33848148 1 ! Hereditary!Sensory!Neuropathy UCEC 0.71 NEFL,DNM2 0.096035489 NTRK1!E>!Trk!receptor!signaling!mediated!by!PI3K!and!PLCEgamma(CCND1,PIK3CA,NRAS,PIK3R1,KRAS);NDRG1!E>!Direct!p53!effectors(VDR,TP53,PIDD,RB1,PTEN,CREBBP);NTRK1!E>!trka!receptor!signaling!pathway(PIK3CA,PIK3R1);NTRK1!E>!p73!transcription!factor!network(MYC,RNF43,RB1,WWOX);NTRK1!E>!p75(NTR)Emediated!signaling(PIK3CA,IRAK1,OMG,TP53,PIK3R1);DNM2!E>!PAR1Emediated!thrombin!signaling!events(PIK3CA,GNAQ,PIK3R1,DNM2);PMP22!E>!a6b1!and!a6b4!Integrin!signaling(ERBB2,PIK3CA,PIK3R1,ERBB3)0.130354569 1 1 ! Li!Fraumeni!and!Related!Syndromes UCEC 0.63 TP53 1.96EE18 TP53!E>!chaperones!modulate!interferon!signaling!pathway(TP53,RB1);TP53!E>!Direct!p53!effectors(VDR,TP53,PIDD,RB1,PTEN,CREBBP);TP53!E>!LKB1!signaling!events(MYC,TP53,ESR1);TP53,CHEK2!E>!cell!cycle:!g2/m!checkpoint(TP53,MYT1);TP53!E>!estrogen!responsive!protein!efp!controls!cell!cycle!and!breast!tumors!growth(TP53,ESR1);TP53!E>!overview!of!telomerase!protein!component!gene!htert!transcriptional!regulation(MYC,TERT,TP53,MZF1,ESR1);CDKN2A!E>!Coregulation!of!Androgen!receptor!activity(CTNNB1,CCND1,CASP8,NKX3E1);CDKN2A,TP53!E>!Hypoxic!and!oxygen!homeostasis!regulation!of!HIFE1Ealpha(TP53,ARNT);TP53!E>!telomeres!telomerase!cellular!aging!and!immortality(MYC,TERT,TP53,RB1,KRAS);CDKN2A!E>!Regulation!of!nuclear!beta!catenin!signaling!and!target!gene!transcription(CTNNB1,CCND1,MYC,TERT);TP53!E>!btg!family!proteins!and!cell!cycle!regulation(CCND1,TP53,RB1);TP53!E>!Transcriptional!!activation!of!!cell!cycle!inhibitor!p21(TP53);TP53,CHEK2!E>!PLK3!signaling!events(CCNE1,TP53);TP53!E>!p53!signaling!pathway(CCND1,CCNE1,TP53,0.458169296 0.57574242 1 ! Lipoprotein!Deficiencies UCEC 1 1 0.064216795 MTTP,APOB,LCAT,SAR1B,APOA1!E>!A1BG(1.19eE02);MTTP,APOB,LCAT,SAR1B,APOA1!E>!SLC27A5(7.36eE03);MTTP,APOB,LCAT,SAR1B,APOA1!E>!HPD(1.19eE02);MTTP,APOB,LCAT,SAR1B,APOA1!E>!ATRN(2.26eE02);MTTP,APOB,LCAT,SAR1B,APOA1,ABCA1!E>!NEU4(1.39eE02)1 1 !

122

Supplementary Table 5: ADAMS results for comorbidity with acute exacerbations of myasthenia gravis. Related to section 5.3.5.

Have& Have& disease& disease& (case& (control& Incidence& set,&204& Incidence& set,&2582& (control& Odds& ICD$9&description patients) (case&set) patients) set) ratio p$value Carpal&tunnel&syndrome 9 0.044118 2 7.75E$04 56.96 2.47E$09 Urinary&tract&infection&site¬&specified 26 0.127451 76 0.029435 4.33 5.95E$09 Pneumonitis&due&to&inhalation&of&food&or&vomitus15 0.073529 21 0.008133 9.041 7.94E$09 Anxiety&state&unspecified 14 0.068627 22 0.008521 8.054 7.39E$08 Unspecified&essential&hypertension 53 0.259804 305 0.118125 2.199 9.73E$08 Esophageal&reflux 19 0.093137 56 0.021689 4.294 8.60E$07 Unspecified&pleural&effusion 12 0.058824 21 0.008133 7.232 1.55E$06 Friedlander\'s&bacillus&infection&in&conditions&classified&elsewhere&and&of&unspecified&site6 0.029412 2 7.75E$04 37.97 3.55E$06 Thyrotoxicosis&without&goiter&or&other&cause&and&without&thyrotoxic&crisis&or&storm6 0.029412 2 7.75E$04 37.97 3.55E$06 Atrial&fibrillation 15 0.073529 41 0.015879 4.631 6.17E$06 Other&specified&disorders&of&pancreatic&internal&secretion4 0.019608 0 0 $1 2.80E$05 Other&specified&idiopathic&peripheral&neuropathy4 0.019608 0 0 $1 2.80E$05 Pure&hypercholesterolemia 20 0.098039 83 0.032146 3.05 3.49E$05 Hemorrhage&complicating&a&procedure 5 0.02451 2 7.75E$04 31.64 3.73E$05 Long$term&(current)&use&of&steroids 5 0.02451 3 0.001162 21.09 9.37E$05 Personal&history&of&noncompliance&with&medical&treatment&presenting&hazards&to&health7 0.034314 13 0.005035 6.815 3.47E$04 Hematoma&complicating&a&procedure 4 0.019608 2 7.75E$04 25.31 3.73E$04 Adrenal&cortical&steroids&causing&adverse&effects&in&therapeutic&use5 0.02451 5 0.001936 12.66 3.73E$04 Nontoxic&uninodular&goiter 3 0.014706 0 0 $1 3.87E$04 Chronic&lymphocytic&thyroiditis 3 0.014706 0 0 $1 3.87E$04 Personal&history&of&malignant&neoplasm&of&bladder3 0.014706 0 0 $1 3.87E$04 Personal&history&of&malignant&neoplasm&of&other&endocrine&glands&and&related&structures3 0.014706 0 0 $1 3.87E$04 Embolism&and&thrombosis&of&other&specified&veins4 0.019608 3 0.001162 16.88 8.20E$04 Depressive&disorder¬&elsewhere&classified 14 0.068627 62 0.024012 2.858 9.55E$04 Toxic&diffuse&goiter&without&thyrotoxic&crisis&or&storm3 0.014706 1 3.87E$04 37.97 0.001465 Unspecified&idiopathic&peripheral&neuropathy 3 0.014706 1 3.87E$04 37.97 0.001465 Unspecified&disorder&of&optic&nerve&and&visual&pathways3 0.014706 1 3.87E$04 37.97 0.001465 Personal&history&of&tobacco&use 8 0.039216 24 0.009295 4.219 0.001632 Diabetes&mellitus&without&complication&type&i¬&stated&as&uncontrolled8 0.039216 25 0.009682 4.05 0.002022 Hypertrophy&(benign)&of&prostate&without&urinary&obstruction5 0.02451 9 0.003486 7.032 0.002323 Bipolar&disorder,&unspecified 4 0.019608 5 0.001936 10.13 0.002624 Unspecified&disorder&of&thyroid 3 0.014706 2 7.75E$04 18.99 0.003465 Retention&of&urine&unspecified 3 0.014706 2 7.75E$04 18.99 0.003465 Other&specified&retention&of&urine 3 0.014706 2 7.75E$04 18.99 0.003465 Migraine&unspecified&without&mention&of&intractable&migraine&without&mention&of&status&migrainosus4 0.019608 6 0.002324 8.438 0.004125 Other&pulmonary&embolism&and&infarction 4 0.019608 6 0.002324 8.438 0.004125 Unspecified&sleep&apnea 4 0.019608 6 0.002324 8.438 0.004125 Anemia&unspecified 11 0.053922 53 0.020527 2.627 0.005828 Methicillin&susceptible&staphylococcus&aureus 3 0.014706 3 0.001162 12.66 0.006558 Obstructive&sleep&apnea&(adult)(pediatric) 3 0.014706 3 0.001162 12.66 0.006558 Tracheostomy&status 4 0.019608 10 0.003873 5.063 0.015584

123

8 REFERENCES

Adameyko, Igor, Francois Lallemend, Alessandro Furlan, Nikolay Zinin, Sergi Aranda, Satish Srinivas Kitambi, Albert Blanchart, et al. 2012. “Sox2 and Mitf Cross-Regulatory Interactions Consolidate Progenitor and Melanocyte Lineages in the Cranial Neural Crest.” Development (Cambridge, England) 139 (2): 397–410. doi:10.1242/dev.065581. http://dev.biologists.org/content/139/2/397.long.

Akavia, Uri David, Oren Litvin, Jessica Kim, Felix Sanchez-Garcia, Dylan Kotliar, Helen C Causton, Panisa Pochanard, Eyal Mozes, Levi a Garraway, and Dana Pe’er. 2010. “An Integrated Approach to Uncover Drivers of Cancer.” Cell 143 (6). Elsevier Inc.: 1005– 17. doi:10.1016/j.cell.2010.11.013. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3013278&tool=pmcentr ez&rendertype=abstract.

Akbani, Rehan, Patrick Kwok Shing Ng, Henrica M J Werner, Maria Shahmoradgoli, Fan Zhang, Zhenlin Ju, Wenbin Liu, et al. 2014. “A Pan-Cancer Proteomic Perspective on The Cancer Genome Atlas.” Nature Communications 5 (May): 3887. doi:10.1038/ncomms4887. http://www.ncbi.nlm.nih.gov/pubmed/24871328.

Anderson, J E. 1983. “Seasonality of Symptomatic Bacterial Urinary Infections in Women.” Journal of Epidemiology and Community Health 37 (4): 286–90. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1052926&tool=pmcentr ez&rendertype=abstract.

Armitage, P, and R Doll. 1954. “The Age Distribution of Cancer and a Multi-Stage Theory of Carcinogenesis.” British Journal of Cancer VIII (I): 1–12.

Arnesen, T, D Gromyko, F Pendino, A Ryningen, J E Varhaug, and J R Lillehaug. 2006. “Induction of Apoptosis in Human Cells by RNAi-Mediated Knockdown of hARD1 and NATH, Components of the Protein N-Alpha-Acetyltransferase Complex.” Oncogene 25 (31): 4350–60. doi:10.1038/sj.onc.1209469. http://dx.doi.org/10.1038/sj.onc.1209469.

Aydin, Iraz T., Rachel D. Melamed, Sarah J. Adams, Mireia Castillo-Martin, Ahu Demir, Diana Bryk, Georg Brunner, et al. 2014. “FBXW7 Mutations in Melanoma and a New Therapeutic Paradigm.” Journal of the National Cancer Institute 106 (6).

Benjamini, Y, and Y Hochberg. 1995. “Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing.” Journal of the Royal Statistical Society. Series B …. http://www.jstor.org/stable/2346101.

Beroukhim, Rameen, Gad Getz, Leia Nghiemphu, Jordi Barretina, Teli Hsueh, David Linhart, Igor Vivanco, et al. 2007. “Assessing the Significance of Chromosomal Aberrations in Cancer: Methodology and Application to Glioma.” Proceedings of the National Academy of Sciences of the United States of America 104 (50): 20007–12. doi:10.1073/pnas.0710052104. 124

http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2148413&tool=pmcentr ez&rendertype=abstract.

Blair, David R., Christopher S. Lyttle, Jonathan M. Mortensen, Charles F. Bearden, Anders Boeck Jensen, Hossein Khiabanian, Rachel Melamed, et al. 2013. “A Nondegenerate Code of Deleterious Variants in Mendelian Loci Contributes to Complex Disease Risk.” Cell 155 (1). Elsevier: 70–80. doi:10.1016/j.cell.2013.08.030. http://www.cell.com/fulltext/S0092-8674(13)01024-6.

Bobak, M. 2001. “The Seasonality of Live Birth Is Strongly Influenced by Socio-Demographic Factors.” Human Reproduction 16 (7): 1512–17. doi:10.1093/humrep/16.7.1512. http://humrep.oxfordjournals.org/content/16/7/1512.long.

Boca, Simina M, Kenneth W Kinzler, Victor E Velculescu, Bert Vogelstein, and Giovanni Parmigiani. 2010. “Patient-Oriented Gene Set Analysis for Cancer Mutation Data.” Genome Biology 11 (11). BioMed Central Ltd: R112. doi:10.1186/gb-2010-11-11-r112. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3156951&tool=pmcentr ez&rendertype=abstract.

Bourne, T David, and David Schiff. 2010. “Update on Molecular Findings, Management and Outcome in Low-Grade Gliomas.” Nature Reviews. Neurology 6 (12). Nature Publishing Group: 695–701. doi:10.1038/nrneurol.2010.159. http://www.ncbi.nlm.nih.gov/pubmed/21045797.

Bowman, A. W., and A. Azzalini. 1997. Applied Smoothing Techniques for Data Analysis. New York: Oxford University Press.

Brauer, Patrick M, and Angela L Tyner. 2010. “Building a Better Understanding of the Intracellular Tyrosine Kinase PTK6 - BRK by BRK.” Biochimica et Biophysica Acta 1806 (1): 66–73. doi:10.1016/j.bbcan.2010.02.003. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2885473&tool=pmcentr ez&rendertype=abstract.

Brennan, Cameron W, Roel G W Verhaak, Aaron McKenna, Benito Campos, Houtan Noushmehr, Sofie R Salama, Siyuan Zheng, et al. 2013. “The Somatic Genomic Landscape of Glioblastoma.” Cell 155 (2): 462–77. doi:10.1016/j.cell.2013.09.034. http://www.ncbi.nlm.nih.gov/pubmed/24120142.

Brinkman, Ryan R, Marie-Pierre Dubé, Guy A Rouleau, Andrew C Orr, and Mark E Samuels. 2006. “Human Monogenic Disorders - a Source of Novel Drug Targets.” Nature Reviews. Genetics 7 (4): 249–60. doi:10.1038/nrg1828.

Brown, G, C Wallin, T Tatusova, K Pruitt, and D Maglott. 2005. “Gene Help: Integrated Access to Genes of Genomes in the Reference Sequence Collection.” Bethesda, MD: National Center for Biotechnology Information.

Bunt, Jens, Talitha G de Haas, Nancy E Hasselt, Danny A Zwijnenburg, Jan Koster, Rogier Versteeg, and Marcel Kool. 2010. “Regulation of Cell Cycle Genes and Induction of 125

Senescence by Overexpression of OTX2 in Medulloblastoma Cell Lines.” Molecular Cancer Research : MCR 8 (10): 1344–57. doi:10.1158/1541-7786.MCR-09-0546. http://mcr.aacrjournals.org/content/8/10/1344.long.

Burns, Jane C, Daniel R Cayan, Garrick Tong, Emelia V Bainto, Christena L Turner, Hiroko Shike, Tomisaku Kawasaki, Yosikazu Nakamura, Mayumi Yashiro, and Hiroshi Yanagawa. 2005. “Seasonality and Temporal Clustering of Kawasaki Syndrome.” Epidemiology (Cambridge, Mass.) 16 (2): 220–25. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2894624&tool=pmcentr ez&rendertype=abstract.

Cao, Juxiang, Lixin Wan, Elke Hacker, Xiangpeng Dai, Stefania Lenna, Celia Jimenez- Cervantes, Yongjun Wang, et al. 2013. “MC1R Is a Potent Regulator of PTEN after UV Exposure in Melanocytes.” Molecular Cell 51 (4): 409–22. doi:10.1016/j.molcel.2013.08.010. http://www.sciencedirect.com/science/article/pii/S1097276513005832.

Cerami, Ethan, Emek Demir, Nikolaus Schultz, Barry S Taylor, and Chris Sander. 2010. “Automated Network Analysis Identifies Core Pathways in Glioblastoma.” PloS One 5 (2): e8918. doi:10.1371/journal.pone.0008918. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2820542&tool=pmcentr ez&rendertype=abstract.

Challa-Malladi, Madhavi, Yen K Lieu, Olivia Califano, Antony B Holmes, Govind Bhagat, Vundavalli V Murty, David Dominguez-Sola, Laura Pasqualucci, and Riccardo Dalla- Favera. 2011. “Combined Genetic Inactivation of β2-Microglobulin and CD58 Reveals Frequent Escape from Immune Recognition in Diffuse Large B Cell Lymphoma.” Cancer Cell 20 (6). Elsevier: 728–40. doi:10.1016/j.ccr.2011.11.006. http://www.cell.com/article/S1535610811004375/fulltext.

Ciriello, Giovanni, Ethan G. Cerami, Chris Sander, and Nikolaus Schultz. 2011. “Mutual Exclusivity Analysis Identifies Oncogenic Network Modules.” Genome Research 22 (2): 398–406. doi:10.1101/gr.125567.111. http://genome.cshlp.org/cgi/content/abstract/22/2/398.

Ciriello, Giovanni, Martin L Miller, Bülent Arman Aksoy, Yasin Senbabaoglu, Nikolaus Schultz, and Chris Sander. 2013. “Emerging Landscape of Oncogenic Signatures across Human Cancers.” Nature Genetics 45 (10). Nature Publishing Group, a division of Macmillan Publishers Limited. All Rights Reserved.: 1127–33. doi:10.1038/ng.2762. http://dx.doi.org/10.1038/ng.2762.

Clement, Virginie, Pilar Sanchez, Nicolas de Tribolet, Ivan Radovanovic, and Ariel Ruiz i Altaba. 2007. “HEDGEHOG-GLI1 Signaling Regulates Human Glioma Growth, Cancer Stem Cell Self-Renewal, and Tumorigenicity.” Current Biology : CB 17 (2): 165–72. doi:10.1016/j.cub.2006.11.033. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1855204&tool=pmcentr ez&rendertype=abstract.

126

Coloma, Preciosa M, Vera E Valkhoff, Giampiero Mazzaglia, Malene Schou Nielsson, Lars Pedersen, Mariam Molokhia, Mees Mosseveld, et al. 2013. “Identification of Acute Myocardial Infarction from Electronic Healthcare Records Using Different Disease Coding Systems: A Validation Study in Three European Countries.” BMJ Open 3 (6). doi:10.1136/bmjopen-2013-002862. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3686251&tool=pmcentr ez&rendertype=abstract.

“Comprehensive Genomic Characterization Defines Human Glioblastoma Genes and Core Pathways.” 2008. Nature 455 (7216): 1061–68. doi:10.1038/nature07385. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2671642&tool=pmcentr ez&rendertype=abstract.

Consortium, The Fantom, Riken Pmi, and Clst Dgt. 2014. “A Promoter-Level Mammalian Expression Atlas.” Nature 507 (7493). Nature Publishing Group: 462–70. doi:10.1038/nature13182. http://www.ncbi.nlm.nih.gov/pubmed/24670764.

Cufi, Perrine, Nadine Dragin, Julia Miriam Weiss, Pilar Martinez-Martinez, Marc H De Baets, Régine Roussin, Elie Fadel, Sonia Berrih-Aknin, and Rozen Le Panse. 2013. “Implication of Double-Stranded RNA Signaling in the Etiology of Autoimmune Myasthenia Gravis.” Annals of Neurology 73 (2): 281–93. doi:10.1002/ana.23791. http://www.ncbi.nlm.nih.gov/pubmed/23280437.

Dai, Mu-Shui, and Hua Lu. 2004. “Inhibition of MDM2-Mediated p53 Ubiquitination and Degradation by Ribosomal Protein L5.” The Journal of Biological Chemistry 279 (43): 44475–82. doi:10.1074/jbc.M403722200. http://www.jbc.org/content/279/43/44475.

Davies, Helen, Graham R Bignell, Charles Cox, Philip Stephens, Sarah Edkins, Sheila Clegg, Jon Teague, et al. 2002. “Mutations of the BRAF Gene in Human Cancer.” Nature 417 (6892): 949–54. doi:10.1038/nature00766. http://www.ncbi.nlm.nih.gov/pubmed/12068308.

DeBerardinis, Ralph J., Nabil Sayed, Dara Ditsworth, and Craig B. Thompson. 2008. “Brick by Brick: Metabolism and Tumor Cell Growth.” Current Opinion in Genetics and Development 18: 54–61. doi:10.1016/j.gde.2008.02.003.

Demokan, Semra, Alice Y Chuang, Kavita M Pattani, David Sidransky, Wayne Koch, and Joseph A Califano. 2014. “Validation of Nucleolar Protein 4 as a Novel Methylated Tumor Suppressor Gene in Head and Neck Cancer.” Oncology Reports 31 (2): 1014–20. doi:10.3892/or.2013.2927. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3896520&tool=pmcentr ez&rendertype=abstract.

Denny, Joshua C, Lisa Bastarache, Marylyn D Ritchie, Robert J Carroll, Raquel Zink, Jonathan D Mosley, Julie R Field, et al. 2013. “Systematic Comparison of Phenome-Wide Association Study of Electronic Medical Record Data and Genome-Wide Association

127

Study Data.” Nature Biotechnology 31 (12). Nature Publishing Group: 1102–11. doi:10.1038/nbt.2749. http://www.ncbi.nlm.nih.gov/pubmed/24270849.

Druker, B J, M Talpaz, D J Resta, B Peng, E Buchdunger, J M Ford, N B Lydon, et al. 2001. “Efficacy and Safety of a Specific Inhibitor of the BCR-ABL Tyrosine Kinase in Chronic Myeloid Leukemia.” The New England Journal of Medicine. Vol. 344. doi:10.1056/NEJM200104053441401.

Eber, Michael R, Michelle Shardell, Marin L Schweizer, Ramanan Laxminarayan, and Eli N Perencevich. 2011. “Seasonal and Temperature-Associated Increases in Gram-Negative Bacterial Bloodstream Infections among Hospitalized Patients.” PloS One 6 (9): e25298. doi:10.1371/journal.pone.0025298. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3180381&tool=pmcentr ez&rendertype=abstract.

Elder, David E. 2010. “Dysplastic Naevi: An Update.” Histopathology.

Falagas, M E, G Peppas, D K Matthaiou, D E Karageorgopoulos, N Karalis, and G Theocharis. 2009. “Effect of Meteorological Variables on the Incidence of Lower Urinary Tract Infections.” European Journal of Clinical Microbiology & Infectious Diseases : Official Publication of the European Society of Clinical Microbiology 28 (6): 709–12. doi:10.1007/s10096-008-0679-z. http://www.ncbi.nlm.nih.gov/pubmed/19104854.

Fine, Jo David, Lorraine B. Johnson, Madeline Weiner, Kuo Ping Li, and Chirayath Suchindran. 2009. “Epidermolysis Bullosa and the Risk of Life-Threatening Cancers: The National EB Registry Experience, 1986-2006.” Journal of the American Academy of Dermatology 60 (2): 203–11.

Finks, JF, NH Osborne, and JD Birkmeyer. 2011. “Trends in Hospital Volume and Operative Mortality for High-Risk Surgery.” New England Journal of … 364: 2128–37. http://www.nejm.org/doi/full/10.1056/NEJMsa1010705#t=articleBackground.

Frankel, Mika, Gabriel Bekö, Michael Timm, Sine Gustavsen, Erik Wind Hansen, and Anne Mette Madsen. 2012. “Seasonal Variations of Indoor Microbial Exposures and Their Relation to Temperature, Relative Humidity, and Air Exchange Rate.” Applied and Environmental Microbiology 78 (23): 8289–97. doi:10.1128/AEM.02069-12. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3497365&tool=pmcentr ez&rendertype=abstract.

Friend, Stephen H, R Bernards, S Rogelj, R Weinberg, JM Rapaport, DM Albert, and TP Dryja. 2014. “A Human DNA Segment with Properties of the Gene That Predisposes to Retinoblastoma and Osteosarcoma.” Accessed June 25. http://www.nature.com/scitable/content/A-human-DNA-segment-with-properties- of-11477.

Gallerani, Massimo, Benedetta Boari, Fabio Manfredini, and Roberto Manfredini. 2011. “Seasonal Variation in Heart Failure Hospitalization.” Clinical Cardiology 34 (6): 389– 94. doi:10.1002/clc.20895. http://www.ncbi.nlm.nih.gov/pubmed/21538387. 128

Goebel, B., Z. Dawy, J. Hagenauer, and J.C. Mueller. “An Approximation to the Distribution of Finite Sample Size Mutual Information Estimates.” IEEE International Conference on Communications, 2005. ICC 2005. 2005 2 (4). Ieee: 1102–6. doi:10.1109/ICC.2005.1494518. http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=1494518.

Gottesman, Omri, Helena Kuivaniemi, Gerard Tromp, W Andrew Faucett, Rongling Li, Teri A Manolio, Saskia C Sanderson, et al. 2013. “The Electronic Medical Records and Genomics (eMERGE) Network: Past, Present, and Future.” Genetics in Medicine : Official Journal of the American College of Medical Genetics 15 (10). American College of Medical Genetics and Genomics: 761–71. doi:10.1038/gim.2013.72. http://dx.doi.org/10.1038/gim.2013.72.

Gudbjartsson, Daniel F, Patrick Sulem, Simon N Stacey, Alisa M Goldstein, Thorunn Rafnar, Bardur Sigurgeirsson, Kristrun R Benediktsdottir, et al. 2008. “ASIP and TYR Pigmentation Variants Associate with Cutaneous Melanoma and Basal Cell Carcinoma.” Nature Genetics 40 (7). Nature Publishing Group: 886–91. doi:10.1038/ng.161. http://dx.doi.org/10.1038/ng.161.

Hanahan, Douglas, and Robert A Weinberg. 2011. “Review Hallmarks of Cancer : The Next Generation.” Cell 144 (5). Elsevier Inc.: 646–74. doi:10.1016/j.cell.2011.02.013. http://dx.doi.org/10.1016/j.cell.2011.02.013.

Hayashi, Reiko, Yuya Goto, Ryuji Ikeda, Kazunari K Yokoyama, and Kenichi Yoshida. 2006. “CDCA4 Is an E2F Transcription Factor Family-Induced Nuclear Factor That Regulates E2F-Dependent Transcriptional Activation and Cell Proliferation.” The Journal of Biological Chemistry 281 (47): 35633–48. doi:10.1074/jbc.M603800200. http://www.ncbi.nlm.nih.gov/pubmed/16984923.

Hoadley, Katherine a., Christina Yau, Denise M. Wolf, Andrew D. Cherniack, David Tamborero, Sam Ng, Max D.M. Leiserson, et al. 2014. “Multiplatform Analysis of 12 Cancer Types Reveals Molecular Classification within and across Tissues of Origin.” Cell 158 (4). Elsevier Inc.: 929–44. doi:10.1016/j.cell.2014.06.049. http://linkinghub.elsevier.com/retrieve/pii/S0092867414008769.

Hoehndorf, Robert, Paul N Schofield, and Georgios V Gkoutos. 2013. “An Integrative, Translational Approach to Understanding Rare and Orphan Genetically Based Diseases.” Interface Focus 3 (2): 20120055. doi:10.1098/rsfs.2012.0055. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3638468&tool=pmcentr ez&rendertype=abstract.

Hoek, Keith S, Natalie C Schlegel, Ossia M Eichhoff, Daniel S Widmer, Christian Praetorius, Steingrimur O Einarsson, Sigridur Valgeirsdottir, et al. 2008. “Novel MITF Targets Identified Using a Two-Step DNA Microarray Strategy.” Pigment Cell & Melanoma Research 21 (6): 665–76. doi:10.1111/j.1755-148X.2008.00505.x. http://www.ncbi.nlm.nih.gov/pubmed/19067971.

129

Hofree, Matan, John P Shen, Hannah Carter, Andrew Gross, and Trey Ideker. 2013. “Network-Based Stratification of Tumor Mutations.” Nature Methods 10 (11): 1108–15. doi:10.1038/nmeth.2651. http://www.ncbi.nlm.nih.gov/pubmed/24037242.

Holmes, Antony B, Alexander Hawson, Feng Liu, Carol Friedman, Hossein Khiabanian, and Raul Rabadan. 2011. “Discovering Disease Associations by Integrating Electronic Clinical Data and Medical Literature.” PloS One 6 (6): e21132. doi:10.1371/journal.pone.0021132. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3121722&tool=pmcentr ez&rendertype=abstract.

Hwang, Harry C, and Bruce E Clurman. 2005. “Cyclin E in Normal and Neoplastic Cell Cycles.” Oncogene 24 (17): 2776–86. doi:10.1038/sj.onc.1208613. http://dx.doi.org/10.1038/sj.onc.1208613.

Kamburov, Atanas, Ulrich Stelzl, Hans Lehrach, and Ralf Herwig. 2013. “The ConsensusPathDB Interaction Database: 2013 Update.” Nucleic Acids Research 41 (Database issue): D793–800. doi:10.1093/nar/gks1055. http://nar.oxfordjournals.org/content/41/D1/D793.

Khiabanian, Hossein, Antony B Holmes, Brendan J Kelly, Mrinalini Gururaj, George Hripcsak, and Raul Rabadan. 2010. “Signs of the 2009 Influenza Pandemic in the New York- Presbyterian Hospital Electronic Health Records.” Edited by Nancy Mock. PloS One 5 (9). Public Library of Science: 8. doi:10.1371/journal.pone.0012658. http://dx.plos.org/10.1371/journal.pone.0012658.

Kishimoto, Masahiro, Takashi Kohno, Koji Okudela, Ayaka Otsuka, Hiroki Sasaki, Chikako Tanabe, Tokuki Sakiyama, et al. 2005. “Mutations and Deletions of the CBP Gene in Human Lung Cancer.” Clinical Cancer Research : An Official Journal of the American Association for Cancer Research 11 (2 Pt 1): 512–19. http://www.ncbi.nlm.nih.gov/pubmed/15701835.

Knudson, AG. 2001. “Two Genetic Hits (more or Less) to Cancer.” Nature Reviews Cancer 1 (November): 637–41. http://www.nature.com/nrc/journal/v1/n2/abs/nrc1101- 157a.html.

Kobayashi, Koichi S, and Peter J van den Elsen. 2012. “NLRC5: A Key Regulator of MHC Class I-Dependent Immune Responses.” Nature Reviews. Immunology 12 (12). Nature Publishing Group, a division of Macmillan Publishers Limited. All Rights Reserved.: 813–20. doi:10.1038/nri3339. http://dx.doi.org/10.1038/nri3339.

Kullback, S. 2012. “Information Theory and Statistics.” Accessed October 21. http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.14.5452.

Lambe, M, P Blomqvist, and R Bellocco. 2003. “Seasonal Variation in the Diagnosis of Cancer: A Study Based on National Cancer Registration in Sweden.” British Journal of Cancer 88 (9). Cancer Research UK: 1358–60. doi:10.1038/sj.bjc.6600901. http://dx.doi.org/10.1038/sj.bjc.6600901. 130

Lambert, G W, C Reid, D M Kaye, G L Jennings, and M D Esler. 2002. “Effect of Sunlight and Season on Serotonin Turnover in the Brain.” Lancet 360 (9348): 1840–42. http://www.ncbi.nlm.nih.gov/pubmed/12480364.

Law, Charity W, Yunshun Chen, Wei Shi, and Gordon K Smyth. 2014. “Voom: Precision Weights Unlock Linear Model Analysis Tools for RNA-Seq Read Counts.” Genome Biology 15 (2): R29. doi:10.1186/gb-2014-15-2-r29. http://genomebiology.com/2014/15/2/R29.

Lawrence, Michael S, Petar Stojanov, Craig H Mermel, James T Robinson, Levi A Garraway, Todd R Golub, Matthew Meyerson, Stacey B Gabriel, Eric S Lander, and Gad Getz. 2014. “Discovery and Saturation Analysis of Cancer Genes across 21 Tumour Types.” Nature 505 (7484). Nature Publishing Group, a division of Macmillan Publishers Limited. All Rights Reserved.: 495–501. doi:10.1038/nature12912. http://dx.doi.org/10.1038/nature12912.

Lawrence, Michael S, Petar Stojanov, Paz Polak, Gregory V Kryukov, Kristian Cibulskis, Andrey Sivachenko, Scott L Carter, et al. 2013. “Mutational Heterogeneity in Cancer and the Search for New Cancer-Associated Genes.” Nature 499 (7457): 214–18. doi:10.1038/nature12213. http://www.ncbi.nlm.nih.gov/pubmed/23770567.

Lee, D-S, J Park, K a Kay, N a Christakis, Z N Oltvai, and A-L Barabási. 2008. “The Implications of Human Metabolic Network Topology for Disease Comorbidity.” Proceedings of the National Academy of Sciences of the United States of America 105 (29): 9880–85. doi:10.1073/pnas.0802208105. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2481357&tool=pmcentr ez&rendertype=abstract.

Lee, Insuk, U Martin Blom, Peggy I Wang, Jung Eun Shim, and Edward M Marcotte. 2011. “Prioritizing Candidate Disease Genes by Network-Based Boosting of Genome-Wide Association Data.” Genome Research 21 (7): 1109–21. doi:10.1101/gr.118992.110. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3129253&tool=pmcentr ez&rendertype=abstract.

Leiserson, Mark D M, Fabio Vandin, Hsin-Ta Wu, Jason R Dobson, Jonathan V Eldridge, Jacob L Thomas, Alexandra Papoutsaki, et al. 2014. “Pan-Cancer Network Analysis Identifies Combinations of Rare Somatic Mutations across Pathways and Protein Complexes.” Nature Genetics, December. Nature Publishing Group. doi:10.1038/ng.3168. http://www.nature.com/doifinder/10.1038/ng.3168.

Leiserson, Mark D. M., Dima Blokh, Roded Sharan, and Benjamin J. Raphael. 2013. “Simultaneous Identification of Multiple Driver Pathways in Cancer.” Edited by Niko Beerenwinkel. PLoS Computational Biology 9 (5): e1003054. doi:10.1371/journal.pcbi.1003054. http://dx.plos.org/10.1371/journal.pcbi.1003054.

Levine, Pamela J, Miriam R Elman, Ravina Kullar, John M Townes, David T Bearden, Rowena Vilches-Tran, Ian McClellan, and Jessina C McGregor. 2013. “Use of Electronic Health Record Data to Identify Skin and Soft Tissue Infections in Primary Care Settings: A 131

Validation Study.” BMC Infectious Diseases 13 (January): 171. doi:10.1186/1471-2334- 13-171. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3637223&tool=pmcentr ez&rendertype=abstract.

Levy, Carmit, Mehdi Khaled, and D.E. David E Fisher. 2006. “MITF: Master Regulator of Melanocyte Development and Melanoma Oncogene.” Trends in Molecular Medicine 12 (9). Elsevier: 406–14. doi:10.1016/j.molmed.2006.07.008. http://linkinghub.elsevier.com/retrieve/pii/S1471491406001699.

Li, Bo, and Colin N Dewey. 2011. “RSEM: Accurate Transcript Quantification from RNA-Seq Data with or without a Reference Genome.” BMC Bioinformatics 12 (January): 323. doi:10.1186/1471-2105-12-323. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3163565&tool=pmcentr ez&rendertype=abstract.

Li, Caiyun G, and Michael R Eccles. 2012. “PAX Genes in Cancer; Friends or Foes?” Frontiers in Genetics 3 (January). Frontiers: 6. doi:10.3389/fgene.2012.00006. http://journal.frontiersin.org/Journal/10.3389/fgene.2012.00006/abstract.

Li, Heng, and Richard Durbin. 2009. “Fast and Accurate Short Read Alignment with Burrows-Wheeler Transform.” Bioinformatics (Oxford, England) 25 (14): 1754–60. doi:10.1093/bioinformatics/btp324. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2705234&tool=pmcentr ez&rendertype=abstract.

Li, Ruwei, Bo Wan, Jun Zhou, Yingli Wang, Ting Luo, Xiuting Gu, Fang Chen, and Long Yu. 2012. “APC/C(Cdh1) Targets Brain-Specific Kinase 2 (BRSK2) for Degradation via the Ubiquitin-Proteasome Pathway.” Edited by Deanna M. Koepp. PloS One 7 (9). Public Library of Science: e45932. doi:10.1371/journal.pone.0045932. http://dx.plos.org/10.1371/journal.pone.0045932.

Llaverias, Gemma, Christiane Danilo, Isabelle Mercier, Kristin Daumer, Franco Capozza, Terence M. Williams, Federica Sotgia, Michael P. Lisanti, and Philippe G. Frank. 2011. “Role of Cholesterol in the Development and Progression of Breast Cancer.” American Journal of Pathology 178 (1): 402–12.

Lomb, N. R. 1976. “Least-Squares Frequency Analysis of Unequally Spaced Data.” Astrophysics and Space Science 39 (2): 447–62. doi:10.1007/BF00648343. http://link.springer.com/10.1007/BF00648343.

Love, Michael I, Wolfgang Huber, and Simon Anders. 2014. “Moderated Estimation of Fold Change and Dispersion for RNA-Seq Data with DESeq2.” Genome Biology 15 (12): 550. doi:10.1186/s13059-014-0550-8. http://genomebiology.com/2014/15/12/550.

Malkin, D, F P Li, L C Strong, J F Fraumeni, C E Nelson, D H Kim, J Kassel, M A Gryka, F Z Bischoff, and M A Tainsky. 1990. “Germ Line p53 Mutations in a Familial Syndrome of

132

Breast Cancer, Sarcomas, and Other Neoplasms.” Science (New York, N.Y.) 250 (4985): 1233–38. http://www.ncbi.nlm.nih.gov/pubmed/1978757.

Manfredi, James J. 2010. “The Mdm2-p53 Relationship Evolves: Mdm2 Swings Both Ways as an Oncogene and a Tumor Suppressor.” Genes & Development 24 (15): 1580–89. doi:10.1101/gad.1941710. http://genesdev.cshlp.org/content/24/15/1580.long.

Margolin, Adam A, Ilya Nemenman, Katia Basso, Chris Wiggins, Gustavo Stolovitzky, Riccardo Dalla Favera, and Andrea Califano. 2006. “ARACNE: An Algorithm for the Reconstruction of Gene Regulatory Networks in a Mammalian Cellular Context.” BMC Bioinformatics 7 Suppl 1 (Suppl 1): S7. doi:10.1186/1471-2105-7-S1-S7. http://www.ncbi.nlm.nih.gov/pubmed/16723010.

Maslov, Sergei, and Kim Sneppen. 2002. “Specificity and Stability in Topology of Protein Networks.” Science 296 (5569): 910–13. doi:10.1126/science.1065103. http://www.ncbi.nlm.nih.gov/pubmed/11988575.

McCarty, Catherine A, Rex L Chisholm, Christopher G Chute, Iftikhar J Kullo, Gail P Jarvik, Eric B Larson, Rongling Li, et al. 2011. “The eMERGE Network: A Consortium of Biorepositories Linked to Electronic Medical Records Data for Conducting Genomic Studies.” BMC Medical Genomics.

McKusick-Nathans Institute of Genetic Medicine, Johns Hopkins University (Baltimore, MD). “Online Mendelian Inheritance in Man, OMIM®.” Http://omim.org/.

Melamed, Rachel D, Hossein Khiabanian, and Raul Rabadan. 2014. “Data-Driven Discovery of Seasonally Linked Diseases from an Electronic Health Records System.” BMC Bioinformatics 15 Suppl 6 (Suppl 6). BioMed Central Ltd: S3. doi:10.1186/1471-2105- 15-S6-S3. http://www.ncbi.nlm.nih.gov/pubmed/25078762.

Mermel, Craig H, Steven E Schumacher, Barbara Hill, Matthew L Meyerson, Rameen Beroukhim, and Gad Getz. 2011. “GISTIC2.0 Facilitates Sensitive and Confident Localization of the Targets of Focal Somatic Copy-Number Alteration in Human Cancers.” Genome Biology 12 (4): R41. doi:10.1186/gb-2011-12-4-r41. http://genomebiology.com/2011/12/4/R41.

Metral, Sylvain, Beata Machnicka, Sylvain Bigot, Yves Colin, Didier Dhermy, and Marie- Christine Lecomte. 2009. “AlphaII-Spectrin Is Critical for Cell Adhesion and Cell Cycle.” The Journal of Biological Chemistry 284 (4): 2409–18. doi:10.1074/jbc.M801324200.

Michaloglou, Chrysiis, Liesbeth C W Vredeveld, Maria S Soengas, Christophe Denoyelle, Thomas Kuilman, Chantal M A M van der Horst, Donné M Majoor, Jerry W Shay, Wolter J Mooi, and Daniel S Peeper. 2005. “BRAFE600-Associated Senescence-like Cell Cycle Arrest of Human Naevi.” Nature 436 (7051): 720–24. doi:10.1038/nature03890. http://www.ncbi.nlm.nih.gov/pubmed/16079850.

Miller, Christopher A, Stephen H Settle, Erik P Sulman, Kenneth D Aldape, and Aleksandar Milosavljevic. 2011. “Discovering Functional Modules by Identifying Recurrent and 133

Mutually Exclusive Mutational Patterns in Tumors.” BMC Medical Genomics 4 (1): 34. doi:10.1186/1755-8794-4-34. http://www.biomedcentral.com/1755-8794/4/34.

Miller, R W, and J H Rubinstein. 1995. “Tumors in Rubinstein-Taybi Syndrome.” American Journal of Medical Genetics 56 (1): 112–15. doi:10.1002/ajmg.1320560125. http://www.ncbi.nlm.nih.gov/pubmed/7747773.

Mo, Qianxing, Sijian Wang, Venkatraman E Seshan, Adam B Olshen, Nikolaus Schultz, Chris Sander, R Scott Powers, Marc Ladanyi, and Ronglai Shen. 2013. “Pattern Discovery and Cancer Gene Identification in Integrated Cancer Genomic Data.” Proceedings of the National Academy of Sciences of the United States of America 110 (11): 4245–50. doi:10.1073/pnas.1208949110. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3600490&tool=pmcentr ez&rendertype=abstract.

Molendijk, Marc L, Judith P M Haffmans, Boudewijn a a Bus, Philip Spinhoven, Brenda W J H Penninx, Jos Prickaerts, Richard C Oude Voshaar, and Bernet M Elzinga. 2012. “Serum BDNF Concentrations Show Strong Seasonal Variation and Correlations with the Amount of Ambient Sunlight.” PloS One 7 (11): e48046. doi:10.1371/journal.pone.0048046. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3487856&tool=pmcentr ez&rendertype=abstract.

Mullighan, Charles G, Jinghui Zhang, Lawryn H Kasper, Stephanie Lerach, Debbie Payne- Turner, Letha A Phillips, Sue L Heatley, et al. 2011. “CREBBP Mutations in Relapsed Acute Lymphoblastic Leukaemia.” Nature 471 (7337). Nature Publishing Group, a division of Macmillan Publishers Limited. All Rights Reserved.: 235–39. doi:10.1038/nature09727. http://dx.doi.org/10.1038/nature09727.

Noushmehr, Houtan, Daniel J Weisenberger, Kristin Diefes, Heidi S Phillips, Kanan Pujara, Benjamin P Berman, Fei Pan, et al. 2010. “Identification of a CpG Island Methylator Phenotype That Defines a Distinct Subgroup of Glioma.” Cancer Cell 17 (5). Elsevier Ltd: 510–22. doi:10.1016/j.ccr.2010.03.017. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2872684&tool=pmcentr ez&rendertype=abstract.

Nowell, P, and D Hungerford. 1960. “A Minute Chromosome in Human Chronic Granulocytic Leukemia.” Science 132: 1497.

Oberg, C, J Li, A Pauley, E Wolf, M Gurney, and U Lendahl. 2001. “The Notch Intracellular Domain Is Ubiquitinated and Negatively Regulated by the Mammalian Sel-10 Homolog.” The Journal of Biological Chemistry 276 (38): 35847–53. doi:10.1074/jbc.M103992200.

Onyango, Patrick, and Andrew P Feinberg. 2011. “A Nucleolar Protein, H19 Opposite Tumor Suppressor (HOTS), Is a Tumor Growth Inhibitor Encoded by a Human Imprinted H19 Antisense Transcript.” Proceedings of the National Academy of Sciences of the United States of America 108 (40): 16759–64. doi:10.1073/pnas.1110904108. 134

http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3189046&tool=pmcentr ez&rendertype=abstract.

Palmieri, Giuseppe, Mariaelena Capone, Maria Ascierto, Giusy Gentilcore, David F Stroncek, Milena Casula, Maria Sini, Marco Palla, Nicola Mozzillo, and Paolo a Ascierto. 2009. “Main Roads to Melanoma.” Journal of Translational Medicine 7 (1): 86. doi:10.1186/1479-5876-7-86. http://www.translational- medicine.com/content/7/1/86.

Park, Juyong, Deok-Sun Lee, Nicholas a Christakis, and Albert-László Barabási. 2009. “The Impact of Cellular Networks on Disease Comorbidity.” Molecular Systems Biology 5 (262): 262. doi:10.1038/msb.2009.16. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2683720&tool=pmcentr ez&rendertype=abstract.

Pasqualucci, Laura, David Dominguez-Sola, Annalisa Chiarenza, Giulia Fabbri, Adina Grunn, Vladimir Trifonov, Lawryn H Kasper, et al. 2011. “Inactivating Mutations of Acetyltransferase Genes in B-Cell Lymphoma.” Nature 471 (7337). Nature Publishing Group: 189–95. doi:10.1038/nature09730. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3271441&tool=pmcentr ez&rendertype=abstract.

Pei, Lei, Pingfang Xie, Enxiang Zhou, Qian Yang, Yi Luo, and Zhonghua Tang. “Overexpression of DEP Domain Containing mTOR-Interacting Protein Correlates with Poor Prognosis in Differentiated Thyroid Carcinoma.” Molecular Medicine Reports 4 (5): 817–23. doi:10.3892/mmr.2011.503. http://www.ncbi.nlm.nih.gov/pubmed/21643629.

Perencevich, Eli N., Jessina C. McGregor, Michelle Shardell, Jon P. Furuno, Anthony D. Harris, J. Glenn Morris, Jr., David N. Fisman, and Judith A. Johnson. 2008. “Summer Peaks in the Incidences of Gram-Negative Bacterial Infection Among Hospitalized Patients •.” Infection Control and Hospital Epidemiology 29 (12): 1124–31. doi:10.1086/592698. http://www.jstor.org/stable/10.1086/592698.

Poulikakos, Poulikos I., and Neal Rosen. 2011. “Mutant BRAF Melanomas—Dependence and Resistance.” Cancer Cell 19 (1). Elsevier Inc.: 11–15. doi:10.1016/j.ccr.2011.01.008. http://www.ncbi.nlm.nih.gov/pubmed/21251612.

Press, WH. 1992. “Numerical Recipes in C : The Art of Scientific Computing.” In , 2nd ed. Cambridge ; New York: Cambridge University Press.

Querol, Luis, and Isabel Illa. 2013. “Myasthenia Gravis and the Neuromuscular Junction.” Current Opinion in Neurology 26 (5): 459–65. doi:10.1097/WCO.0b013e328364c079. http://www.ncbi.nlm.nih.gov/pubmed/23945282.

Raimondi, Sara, Francesco Sera, Sara Gandini, Simona Iodice, Saverio Caini, Patrick Maisonneuve, and Maria Concetta Fargnoli. 2008. “MC1R Variants, Melanoma and Red Hair Color Phenotype: A Meta-Analysis.” International Journal of Cancer. Journal

135

International Du Cancer 122 (12): 2753–60. doi:10.1002/ijc.23396. http://www.ncbi.nlm.nih.gov/pubmed/18366057.

Remke, Marc, Vijay Ramaswamy, John Peacock, David J H Shih, Christian Koelsche, Paul a Northcott, Nadia Hill, et al. 2013. “TERT Promoter Mutations Are Highly Recurrent in SHH Subgroup Medulloblastoma.” Acta Neuropathologica 126 (6): 917–29. doi:10.1007/s00401-013-1198-2. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3830749&tool=pmcentr ez&rendertype=abstract.

Rodó, Xavier, Joan Ballester, Dan Cayan, Marian E Melish, Yoshikazu Nakamura, Ritei Uehara, and Jane C Burns. 2011. “Association of Kawasaki Disease with Tropospheric Wind Patterns.” Scientific Reports 1 (January). Nature Publishing Group: 152. doi:10.1038/srep00152. http://www.nature.com/srep/2011/111110/srep00152/full/srep00152.html.

Sato, S, K Roberts, G Gambino, A Cook, T Kouzarides, and C R Goding. 1997. “CBP/p300 as a Co-Factor for the Microphthalmia Transcription Factor.” Oncogene 14 (25): 3083–92. doi:10.1038/sj.onc.1201298. http://www.ncbi.nlm.nih.gov/pubmed/9223672.

Scargle, Jeffrey D. 1982. “Statistical Aspects of Spectral Analysis of Unevenly Spaced Data.” The Astrophysical Journal 263: 835–53. http://articles.adsabs.harvard.edu/cgi- bin/nph- iarticle_query?bibcode=1982ApJ...263..835S&db_key=AST&page_ind=0&data_type=GI F&type=SCREEN_VIEW&classic=YES.

Schwienbacher, C, L Gramantieri, R Scelfo, a Veronese, G a Calin, L Bolondi, C M Croce, G Barbanti-Brodano, and M Negrini. 2000. “Gain of Imprinting at Chromosome 11p15: A Pathogenetic Mechanism Identified in Human Hepatocarcinomas.” Proceedings of the National Academy of Sciences of the United States of America 97 (10): 5445–49. doi:10.1073/pnas.090087497. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=25848&tool=pmcentrez &rendertype=abstract.

Sedgewick, Andrew J, Stephen C Benz, Shahrooz Rabizadeh, Patrick Soon-Shiong, and Charles J Vaske. 2013. “Learning Subgroup-Specific Regulatory Interactions and Regulator Independence with PARADIGM.” Bioinformatics (Oxford, England) 29 (13): i62–70. doi:10.1093/bioinformatics/btt229. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3694636&tool=pmcentr ez&rendertype=abstract.

Shechter, Ishaiahu, Peihua Dai, Liang Huo, and Guimin Guan. 2003. “IDH1 Gene Transcription Is Sterol Regulated and Activated by SREBP-1a and SREBP-2 in Human Hepatoma HepG2 Cells: Evidence That IDH1 May Regulate Lipogenesis in Hepatic Cells.” Journal of Lipid Research 44 (11): 2169–80. doi:10.1194/jlr.M300285-JLR200.

136

Shrager, J, JM Tenenbaum, and M Travers. 2010. “Cancer Commons: Biomedicine in the Internet Age.” In Collaborative Computational Technologies for Biomedical Research, edited by S Elkins, M Hupcey, and A Williams.

Simka, M. 2010. “Seasonal Variations in the Onset and Healing Rates of Venous Leg Ulcers.” Phlebology / Venous Forum of the Royal Society of Medicine 25 (1): 29–34. doi:10.1258/phleb.2009.008072. http://www.ncbi.nlm.nih.gov/pubmed/20118343.

Smyth, Gordon K. 2004. “Linear Models and Empirical Bayes Methods for Assessing Differential Expression in Microarray Experiments.” Statistical Applications in Genetics and Molecular Biology 3 (January): Article3. doi:10.2202/1544-6115.1027. http://www.ncbi.nlm.nih.gov/pubmed/16646809.

Sokal, R.R., and F.J. Rohif. 1981. Biometry: The Principles and Practice of Statistics in Biological Research. New York: Freeman.

Stark, Chris, Bobby-Joe Breitkreutz, Teresa Reguly, Lorrie Boucher, Ashton Breitkreutz, and Mike Tyers. 2006. “BioGRID: A General Repository for Interaction Datasets.” Nucleic Acids Research 34 (Database issue): D535–39. doi:10.1093/nar/gkj109. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1347471&tool=pmcentr ez&rendertype=abstract.

Stein, Brian D, Adriana Bautista, Glen T Schumock, Todd a Lee, Jeffery T Charbeneau, Diane S Lauderdale, Edward T Naureckas, David O Meltzer, and Jerry a Krishnan. 2012. “The Validity of International Classification of Diseases, Ninth Revision, Clinical Modification Diagnosis Codes for Identifying Patients Hospitalized for COPD Exacerbations.” Chest 141 (1): 87–93. doi:10.1378/chest.11-0024. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3251268&tool=pmcentr ez&rendertype=abstract.

Subramanian, Aravind, Pablo Tamayo, Vamsi K Mootha, Sayan Mukherjee, Benjamin L Ebert, Michael A Gillette, Amanda Paulovich, et al. 2005. “Gene Set Enrichment Analysis: A Knowledge-Based Approach for Interpreting Genome-Wide Expression Profiles.” Proceedings of the National Academy of Sciences of the United States of America 102 (43): 15545–50. doi:10.1073/pnas.0506580102. http://www.pnas.org/cgi/content/abstract/102/43/15545.

Szczurek, Ewa, and Niko Beerenwinkel. 2014. “Modeling Mutual Exclusivity of Cancer Mutations.” PLoS Computational Biology 10 (3): e1003503. doi:10.1371/journal.pcbi.1003503. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3967923&tool=pmcentr ez&rendertype=abstract.

Tang, Ning, John Stein, Renee Y Hsia, Judith H Maselli, and Ralph Gonzales. 2010. “Trends and Characteristics of US Emergency Department Visits, 1997-2007.” JAMA : The Journal of the American Medical Association 304 (6). American Medical Association: 664–70. doi:10.1001/jama.2010.1112. http://jama.jamanetwork.com/article.aspx?articleid=186383&resultClick=3. 137

Taniguchi, Kenichiro, Anoush E Anderson, Ann E Sutherland, and David Wotton. 2012. “Loss of Tgif Function Causes Holoprosencephaly by Disrupting the SHH Signaling Pathway.” PLoS Genetics 8 (2): e1002524. doi:10.1371/journal.pgen.1002524. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3285584&tool=pmcentr ez&rendertype=abstract.

Tanno, Toshihiko, Natarajan V Bhanu, Patricia A Oneal, Sung-Ho Goh, Pamela Staker, Y Terry Lee, John W Moroney, et al. 2007. “High Levels of GDF15 in Thalassemia Suppress Expression of the Iron Regulatory Protein Hepcidin.” Nature Medicine 13 (9). Nature Publishing Group: 1096–1101. doi:10.1038/nm1629. http://dx.doi.org/10.1038/nm1629.

Tarca, Adi Laurentiu, Sorin Draghici, Purvesh Khatri, Sonia S Hassan, Pooja Mittal, Jung-Sun Kim, Chong Jai Kim, Juan Pedro Kusanovic, and Roberto Romero. 2009. “A Novel Signaling Pathway Impact Analysis.” Bioinformatics (Oxford, England) 25 (1): 75–82. doi:10.1093/bioinformatics/btn577. http://bioinformatics.oxfordjournals.org/cgi/content/abstract/25/1/75.

“The Cancer Genome Atlas.” Http://www.cancergenome.nih.gov/.

Therneau, Terry. 2012. “A Package for Survival Analysis in S. R Package Version.”

Tieder, Joel S, Matthew Hall, Katherine a Auger, Paul D Hain, Karen E Jerardi, Angela L Myers, Suraiya S Rahman, Derek J Williams, and Samir S Shah. 2011. “Accuracy of Administrative Billing Codes to Detect Urinary Tract Infection Hospitalizations.” Pediatrics 128 (2): 323–30. doi:10.1542/peds.2010-2064. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3146355&tool=pmcentr ez&rendertype=abstract.

Torkamani, Ali, and Nicholas J Schork. 2009. “Identification of Rare Cancer Driver Mutations by Network Reconstruction.” Genome Research 19 (9): 1570–78. doi:10.1101/gr.092833.109. http://genome.cshlp.org/cgi/content/abstract/19/9/1570.

Torti, Suzy V, and Frank M Torti. 2013. “Iron and Cancer: More Ore to Be Mined.” Nature Reviews. Cancer 13 (5). Nature Publishing Group, a division of Macmillan Publishers Limited. All Rights Reserved.: 342–55. doi:10.1038/nrc3495. http://dx.doi.org/10.1038/nrc3495.

Trifonov, Vladimir, Laura Pasqualucci, Enrico Tiacci, Brunangelo Falini, and Raul Rabadan. 2013. “Statistical Algorithm for Variant Frequency Identification.” BMC Systems Biology In Press.

Turcan, Sevin, Daniel Rohle, Anuj Goenka, Logan A. Walsh, Fang Fang, Emrullah Yilmaz, Carl Campos, et al. 2012. “IDH1 Mutation Is Sufficient to Establish the Glioma Hypermethylator Phenotype.” Nature. doi:10.1038/nature10866.

138

Upshur, Ross E G, Rahim Moineddin, Eric Crighton, Lori Kiefer, and Muhammad Mamdani. 2005. “Simplicity within Complexity: Seasonality and Predictability of Hospital Admissions in the Province of Ontario 1988-2001, a Population-Based Analysis.” BMC Health Services Research 5 (1): 13. doi:10.1186/1472-6963-5-13. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=549216&tool=pmcentre z&rendertype=abstract.

Vandin, Fabio, Eli Upfal, and Benjamin J Raphael. 2011. “Algorithms for Detecting Significantly Mutated Pathways in Cancer.” Journal of Computational Biology : A Journal of Computational Molecular Cell Biology 18 (3): 507–22. doi:10.1089/cmb.2010.0265. http://www.ncbi.nlm.nih.gov/pubmed/21385051.

———. 2012. “De Novo Discovery of Mutated Driver Pathways in Cancer.” Genome Research 22 (2): 375–85. doi:10.1101/gr.120477.111. http://www.ncbi.nlm.nih.gov/pubmed/21653252.

Varadan, Vinay, and Dimitris Anastassiou. 2006. “Inference of Disease-Related Molecular Logic from Systems-Based Microarray Analysis.” PLoS Computational Biology 2 (6): e68. doi:10.1371/journal.pcbi.0020068. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1479089&tool=pmcentr ez&rendertype=abstract.

Vaske, Charles J, Stephen C Benz, J Zachary Sanborn, Dent Earl, Christopher Szeto, Jingchun Zhu, David Haussler, and Joshua M Stuart. 2010. “Inference of Patient-Specific Pathway Activities from Multi-Dimensional Cancer Genomics Data Using PARADIGM.” Bioinformatics (Oxford, England) 26 (12): i237–45. doi:10.1093/bioinformatics/btq182. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2881367&tool=pmcentr ez&rendertype=abstract.

Verhaak, Roel G W, Katherine a Hoadley, Elizabeth Purdom, Victoria Wang, Yuan Qi, Matthew D Wilkerson, C Ryan Miller, et al. 2010. “Integrated Genomic Analysis Identifies Clinically Relevant Subtypes of Glioblastoma Characterized by Abnormalities in PDGFRA, IDH1, EGFR, and NF1.” Cancer Cell 17 (1). Elsevier Ltd: 98–110. doi:10.1016/j.ccr.2009.12.020. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2818769&tool=pmcentr ez&rendertype=abstract.

Vlachos, Adrianna, Philip S Rosenberg, Eva Atsidaftos, Blanche P Alter, and Jeffrey M Lipton. 2012. “Incidence of Neoplasia in Diamond Blackfan Anemia: A Report from the Diamond Blackfan Anemia Registry.” Blood 119 (16): 3815–19. doi:10.1182/blood- 2011-08-375972. http://bloodjournal.hematologylibrary.org/content/119/16/3815?variant=long&sso- checked=1.

Wang, Ping-Yuan, Jian Weng, and Richard G W Anderson. 2005. “OSBP Is a Cholesterol- Regulated Scaffolding Protein in Control of ERK 1/2 Activation.” Science (New York, N.Y.) 307 (December 2004): 1472–76. doi:10.1126/science.1107710. 139

Wang, Xiaoxia, Sherry Towers, Sarada Panchanathan, and Gerardo Chowell. 2013. “A Population Based Study of Seasonality of Skin and Soft Tissue Infections: Implications for the Spread of CA-MRSA.” PloS One 8 (4): e60872. doi:10.1371/journal.pone.0060872. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3614932&tool=pmcentr ez&rendertype=abstract.

Wang, Yingli, Bo Wan, Dawei Li, Jun Zhou, Ruwei Li, Meirong Bai, Fang Chen, and Long Yu. 2012. “BRSK2 Is Regulated by ER Stress in Protein Level and Involved in ER Stress- Induced Apoptosis.” Biochemical and Biophysical Research Communications 423 (4): 813–18. doi:10.1016/j.bbrc.2012.06.046. http://www.sciencedirect.com/science/article/pii/S0006291X12011345.

Watanabe, Satosi. 1960. “Information Theoretical Analysis of Multivariate Correlation.” IBM Journal of Research and Development 4 (January): 66–82. doi:10.1147/rd.41.0066.

Watson, Ian R, Koichi Takahashi, P Andrew Futreal, and Lynda Chin. 2013. “Emerging Patterns of Somatic Mutations in Cancer.” Nature Reviews. Genetics 14 (10). Nature Publishing Group: 703–18. doi:10.1038/nrg3539. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=4014352&tool=pmcentr ez&rendertype=abstract.

Welcker, Markus, Amir Orian, Jianping Jin, Jonathan E Grim, J Wade Harper, Robert N Eisenman, and Bruce E Clurman. 2004. “The Fbw7 Tumor Suppressor Regulates Glycogen Synthase Kinase 3 Phosphorylation-Dependent c-Myc Protein Degradation.” Proceedings of the National Academy of Sciences of the United States of America 101 (24): 9085–90.

Wilkerson, Matthew D, and D Neil Hayes. 2010. “ConsensusClusterPlus: A Class Discovery Tool with Confidence Assessments and Item Tracking.” Bioinformatics (Oxford, England) 26 (12): 1572–73. doi:10.1093/bioinformatics/btq170. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2881355&tool=pmcentr ez&rendertype=abstract.

Winthorst, Wim H, Wendy J Post, Ybe Meesters, Brenda W H J Penninx, and Willem A Nolen. 2011. “Seasonality in Depressive and Anxiety Symptoms among Primary Care Patients and in Patients with Depressive and Anxiety Disorders; Results from the Netherlands Study of Depression and Anxiety.” BMC Psychiatry 11 (January): 198. doi:10.1186/1471-244X-11-198. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3280179&tool=pmcentr ez&rendertype=abstract.

Wu, Di, and Gordon K Smyth. 2012. “Camera: A Competitive Gene Set Test Accounting for Inter-Gene Correlation.” Nucleic Acids Research 40 (17): e133. doi:10.1093/nar/gks461. http://nar.oxfordjournals.org/content/40/17/e133.

Wu, Guanming, Xin Feng, and Lincoln Stein. 2010. “A Human Functional Protein Interaction Network and Its Application to Cancer Data Analysis.” Genome Biology 11 (5): R53. 140

doi:10.1186/gb-2010-11-5-r53. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2898064&tool=pmcentr ez&rendertype=abstract.

Yajima, Ichiro, Mayuko Y Kumasaka, Nguyen Dinh Thang, Yuji Goto, Kozue Takeda, Machiko Iida, Nobutaka Ohgami, et al. 2011. “Molecular Network Associated with MITF in Skin Melanoma Development and Progression.” Journal of Skin Cancer 2011 (January): 730170. doi:10.1155/2011/730170. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3199194&tool=pmcentr ez&rendertype=abstract.

Yamamoto, Yukiya, Akihiro Abe, and Nobuhiko Emi. 2014. “Clarifying the Impact of Polycomb Complex Component Disruption in Human Cancers.” Molecular Cancer Research : MCR 12 (4): 479–84. doi:10.1158/1541-7786.MCR-13-0596. http://mcr.aacrjournals.org/content/12/4/479.long#ref-16.

Yang, Xiang-Jiao. 2004. “The Diverse Superfamily of Lysine Acetyltransferases and Their Roles in Leukemia and Other Diseases.” Nucleic Acids Research 32 (3): 959–76. doi:10.1093/nar/gkh252. http://nar.oxfordjournals.org/content/32/3/959.full.

Zheng, Yu, Zebin Wang, Wenjun Bie, Patrick M Brauer, Bethany E Perez White, Jing Li, Veronique Nogueira, et al. 2013. “PTK6 Activation at the Membrane Regulates Epithelial-Mesenchymal Transition in Prostate Cancer.” Cancer Research 73 (17). American Association for Cancer Research: 5426–37. doi:10.1158/0008-5472.CAN-13- 0443. http://cancerres.aacrjournals.org/content/73/17/5426.full.

Zhu, Haihao, Jaime Acquaviva, Pranatartiharan Ramachandran, Abraham Boskovitz, Steve Woolfenden, Rolf Pfannl, Roderick T Bronson, et al. 2009. “Oncogenic EGFR Signaling Cooperates with Loss of Tumor Suppressor Gene Functions in Gliomagenesis.” Proceedings of the National Academy of Sciences of the United States of America 106 (8): 2712–16. doi:10.1073/pnas.0813314106. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2650331&tool=pmcentr ez&rendertype=abstract.

141