Integrating Omics: the Future of IBD?

Current Role of Immunosuppressants in IBD: Special Lecture

Dig Dis 2014;32(suppl 1):96–102 DOI: 10.1159/000367836

Integrating Omics: The Future of IBD?

Claudio Fiocchi

Department of Gastroenterology and Hepatology, Digestive Disease Institute, and Department of Pathobiology, Lerner Research Institute, The Cleveland Clinic Foundation, Cleveland, Ohio , USA

Key Words plying personalized medicine and far more effective thera- IBD · Crohn’s disease · Ulcerative colitis · Omes · Omics · pies to individual patients with Crohn’s disease and ulcer- Integrome · Interactome ative colitis. For the practicing gastroenterologist, an omics- based delivery of healthcare may be intimidating, but it must be accepted and implemented if he or she is to provide the Abstract best possible care to IBD patients. © 2014 S. Karger AG, Basel The complexity of IBD is well recognized as are the putative four major components of its pathogenesis, i.e. environment, genetic makeup, gut microbiota and mucosal immune response. Each of these components is extremely complex on Introduction its own, and at present should be more appropriately defined by the terms ‘exposome’, ‘genome’, ‘microbiome’ and ‘immu- There are still multiple unanswered issues in regard to nome’, respectively, based on the ‘ome’ suffix that refers to a IBD that range from its etiology to the reasons underlying totality of some sort. None of these ‘omes’ is apparently ca- its variable manifestations, the challenge of a precise diag- pable of causing IBD by itself; it is instead the intricate and nosis, the choice of an ideal therapy, the need for surgery, reciprocal interaction among them, through the so-called the threat of cancer and the prediction of the ultimate out- ‘IBD interactome’, that results in the emergence of IBD, or come. There is, however, one single issue where there is more appropriately the ‘IBD integrome’. To deal with and un- general agreement among investigators, physicians and derstand such overwhelming biological complexity, new ap- patients alike – its overwhelming complexity. In fact, both proaches and tools are needed, and these are represented by forms of IBD, Crohn’s disease (CD) and ulcerative colitis ‘omics’, defined as the study of related sets of biological mol- (UC), are prototypical examples of a class of what are en- ecules in a comprehensive fashion, such as genomics, tran- titled complex diseases, which include many other chron- scriptomics, proteomics, metabolomics, and so on. Numer- ic inflammatory and autoimmune disorders such as asth- ous bioinformatics-based tools are available to explore and ma, rheumatoid arthritis, psoriasis, multiple sclerosis, sys- take advantage of the massive amount of information that temic lupus erythematosus and various forms of allergic can be generated by the analysis of the various omes and reactions. All of these conditions have in common factors their interactions, aiming at identifying the molecular inter- that include exposure to a wide range of poorly defined actome underlying any particular status of health and dis- environmental agents, a genetic predisposition, an abnor- ease. These novel approaches are fully applicable to IBD and mal microbial environment and an aberrant immune re- allow us to achieve the ultimate goal of developing and ap- sponse. These four key factors, their combinations and

© 2014 S. Karger AG, Basel Claudio Fiocchi, MD 0257–2753/14/0327–0096$39.50/0 Department of Pathobiology, Lerner Research Institute The Cleveland Clinic E-Mail [email protected] Cleveland, Ohio 44195 (USA) www.karger.com/ddi E-Mail fiocchc @ ccf.org Downloaded by: C.H.U. de Vigo 198.143.45.65 - 4/7/2016 12:11:31 PM their interactions can be both similar and different in each type of inflammatory or autoimmune disease, and these similarities and differences are involved in the emergence of any particular disease as well as its fate. The environment, the genetic makeup, the gut microbiota and the mu- Exposome Genome cosal immune response are the currently accepted compo- Color version available online nents of IBD pathogenesis [1], and this is likely to be cor- IBD rect. Each one of them is recognized to be extremely complex given the fact that the environment around us is made of countless agents, the genes of several hundred Microbiome Immunome variants or mutations, the gut microbiota of trillions of microorganisms and the immune response of dozens of different cell types producing hundreds of biologically ac- tive molecules. Given this complexity, the word ‘component’, as applied in component of a system, should be re- Fig. 1. The currently accepted four basic components of IBD placed by the word ‘ome’, a Greek-derived suffix that re- pathogenesis represented as overlapping omes: the exposome (en- fers to a totality of some sort. Thus, based on this more vironmental factors), the genome (genetic makeup), the microbi- fitting definition, the four components or ‘omes’ involved ome (gut microbiota) and the immunome (the immune system). in IBD should be respectively called ‘exposome’, ‘genome’, ‘microbiome’ and ‘immunome’ (fig. 1 ).

such as the one by Renz et al. [2] , which envisions more Omes and IBD Pathogenesis than 2 dozen components (omes) and almost as many steps (interactions). Facing this daunting and seemingly One key question related to the above pathogenic qua- overwhelming complexity, it becomes obvious that under- drumvirate is whether all four omes are equally important standing the mechanisms underlying complex conditions in IBD pathogenesis or whether some are more determin- like IBD requires totally new and far more comprehensive ing than others. At the moment there is not enough infor- approaches. This need is exemplarily argued by Fischer [3] : mation to answer this question, but the exposome may ‘To understand complex biological systems it is not enough play a more dominant role given the fact that the genome to characterize the individual molecules in the system. It is cannot have significantly changed in less than a century also necessary to obtain an understanding of the interac- (the period of time when CD and UC appeared), the rec- tions among molecules, particularly in complex diseases. ognition that the gut microbiome is actually part of and With recent technological advances the focus is shifting modulated by the exposome (through diet and xenobiot- toward interpreting data generated by omics technologies, ics), and the assumption that the immunome is nothing that allow to investigate how regulatory processes are dis- more than the consequent effector arm of the intestinal rupted and cause disease.’ In support of this statement is inflammatory response. What is probably correct, how- the comparison between a man-made complex machine, ever, is that no single ome by itself is capable of triggering like a supersonic aircraft, whose advanced performance IBD, and a functional interaction and integration of the depends on innumerous and highly integrated electronic exposome, genome, microbiome and immunome is the circuits, with nature-made but equally if not even more sine qua non condition to develop CD or UC. This forms complex liver cells, which also depend on innumerous and the basis for the concept of an ‘IBD integrome’, and with highly integrated biological circuits [3] . it the need to use ‘omics’ (to be defined later in this review) Another justification for an omics approach to IBD (or as the innovative and unavoidable way to look at IBD. any other complex disorder for the same reason) is the Until recently the traditional paradigm used to explain often forgotten but omnipresent extreme variability of complex diseases included 1 genotype leading to 1 pheno- biological behaviors in humans. Each human being is type and 1 disease, implying 3 components (or 3 omes) and unique and no two alike reactions will ever occur in re- 2 steps (or 2 interactions). With increasing appreciation of sponse to the same identical challenge even in identical biological diversity and complexity in the last couple of de- twins. Let’s take, for example, the response to a tiny and cades, far more intricate paradigms have been proposed, localized proinflammatory stimulus. In a recent study,

IBD Omics Dig Dis 2014;32(suppl 1):96–102 97 DOI: 10.1159/000367836 Downloaded by: C.H.U. de Vigo 198.143.45.65 - 4/7/2016 12:11:31 PM blisters were induced by the injection of an irritant in the skin of healthy subjects, who were then given aspirin to limit inflammation [4] . By measuring the variation in the number of inflammatory cells in the blisters, the subjects could be clearly separated into responders and nonre- sponders, with a reduction or lack of reduction of the in- Exposome Genome Color version available online flammatory infiltrates, respectively [4]. With this seemingly naive but very informative experiment, it becomes clear that a dichotomy exists in regard to the duration and severity of an acute inflammatory response arising from differentially expressed proresolution pathways. When this notion is applied to far more complicated inflammatory processes, such as the one ongoing in the bowel of CD and UC patients, and in a far richer microenviron- ment such as the gut, it is easy to understand why the study of a single ome, no matter in how much depth, can- Microbiome Immunome not provide an answer to the totality of the events resulting from the interaction with all other pathogenically relevant omes. Thus, in light of this evidence two conclu- sions can be reached: (1) IBD is complicated because biology is complicated, and (2) how can we expect to understand or even cure IBD by studying and modulating Fig. 2. The four basic omes of IBD pathogenesis, i.e. the exposome, one single ome at a time, as we currently do? the genome, the microbiome and the immunome, form an IBD integrome, in which each ome interacts with the others through a series of multiple reciprocal positive and negative interactions. Omics: An Approach to Ome Interactions in IBD

Let’s now revisit the previously introduced concept of IBD integrome and look at it taking into account the re- possible interactions becomes staggeringly high. A logical quirement of new approaches to understand IBD and de- conclusion to this line of thinking is that no current single velop better forms of treatments that take into account study, single experiment or single result will ever be able the above-mentioned complexity. The four IBD omes, to provide meaningful answers to the overwhelming in- the exposome, genome, microbiome and immunome, are tricacy of IBD. It also becomes evident that completely all interrelated through multiple and reciprocal interac- new approaches are needed to untangle this intricacy, and tions among them, with every single interaction likely im- perhaps omics are those that can provide meaningful and parting a positive stimulatory signal or a negative sup- practical answers. Omics are defined as the study of re- pressive signal as shown in figure 2 . In this simplistic rep- lated sets of biological molecules in a comprehensive resentation there are 4 omes and 6 sets of positive and fashion [6] , and examples of omics relevant to IBD are negative signals, resulting in 24 possible interactions. genomics, epigenomics, transcriptomics, proteomics and Let’s consider now the composition of each ome by tak- metabolomics, but many more are also important. ing, for example, the genome. The currently known genetic variants associated with both forms of IBD are at least 163 according to the most comprehensive genome- Omics-Based Systems wide association study to date [5] , although more genetic variants are likely to exist. By then replacing the whole How can IBD omics help or, in other words, what kind single genome, which is subject to 12 positive and nega- of practical and useful information can they provide? Let’s tive signals, with 163 the number of interactions with the take, for example, omics obtainable from a human host. genome alone grows to 1,956 (163 × 12). If we apply this Genomics can tell us about alterations in genetic factors, very basic and simplified calculation to the myriad of such as single nucleotide polymorphisms, copy number components in each of the other 3 omes, the number of variations, etc.; transcriptomics can reveal patterns of

98 Dig Dis 2014;32(suppl 1):96–102 Fiocchi DOI: 10.1159/000367836

Downloaded by: C.H.U. de Vigo 198.143.45.65 - 4/7/2016 12:11:31 PM gene expression (activation or repression) in specific tis- associated with the disease [8] . When combined and used sues; proteomics can enumerate what proteins are made at a systemic level, these networks form the basis of the in normal or diseased organs and their relative abundance, human ‘diseasome’ [9] or, if combined and used at an or- and metabolomics can generate profiles of metabolites re- gan level, they can form an organ-specific ome, such as the sulting from processing of a number of proteins and other ‘gutome’ in the case of the gastrointestinal tract [10] . molecules. If we take the gut microbiome, 16S surveys can define which microorganisms are present in the lumen, the mucosa or the stools, metagenomics identifies which Omics-Based Information microbial genes are present in normal and inflamed gut, and metatranscriptomics and metaproteomics can tell us As the sophistication and power of these analytical and what microbes are doing and producing. research tools increases, so are the expectations of those The above methods can generate huge amounts of data that want to exploit their full capabilities. Can omics ap- on the molecules composing the omes, but they do not tell proaches predict who will develop specific diseases? Can us how these molecules or the omes they form interact omics help us make more precise diagnoses? If a disease is among themselves to form the functional ‘molecular in- already established, can omics tell us its final outcome, teractome’ that defines any unique status of health or dis- predict response to therapy and risk of complications? ease. To understand this, it is necessary to introduce the Can omics give us what the best therapies will be or how concepts of ‘nodes’ and ‘edges’ used in systems biology [7] . to modify them during the course of the disease? These are Molecular interactions are ongoing incessantly at all levels all valid and justifiable questions, but omics technologies of life: social, biological, physical and otherwise. Let’s take, have not begun to be utilized to any significant scale in as a simple but representative example, what goes on with human diseases, and certainly not yet in IBD, and they are airline hubs and the flights they control. A large number not sufficiently developed out so that we can choose the of cities have airports and flights that arrive into and de- most appropriate and get the answers we are looking for. part from them, but only a few cities, those where the air- Nevertheless, some initial examples do provide a prag- line hubs are located, have the power of controlling which matic glimpse of the future of omics in IBD and of their and how many flights go where and when. Making an innovative diagnostic and clinical capabilities. Transcrip- analogy in biological terms, these controlling hubs are the tomics can be used to predict response to treatment by ‘nodes’ of the system, and the flights in whatever direction analyzing mucosal gene signatures in CD and UC patients are the ‘edges’. This concept is crucial because it allows [11]. Transcriptomics can reveal that the endoscopically identifying the central regulators or hubs of multiple con- and histologically normal mucosa of UC patients in clini- nections in any biological response, in both physiological cal remission is not really normal, as its gene expression and pathological conditions. This is done by primarily de- pattern is still different from that of the normal mucosa of termining the highest number of edges coming from or healthy controls or uninvolved UC mucosa [12] . Tran- arriving to any particular node, which then becomes the scriptomics can show which genes are downregulated and dominant node or central regulator of the process being which genes remain upregulated in the mucosa of IBD studied. Once the dominant (single or multiple) regulator patients after receiving anti-TNF therapy, and therefore is identified, it is only logical that studies and modulatory allow the identification of a new therapeutic target gene in approaches should be focused on it, and not on periph- nonresponsive subjects [13] . Stool metabolomics can ac- eral and less important nodes of the interactome network. curately differentiate metabolites associated with healthy A number of diverse interactome networks exist in nature controls from those produced in patients with CD, UC or made up of vastly different omes that can provide abun- pouchitis [14] . Breath metabolomics can detect numerous dant information on biological interactions and outcomes significant differences between healthy and IBD individu- [8]. Just to cite a few, in a ‘protein interaction network’, als [unpubl. observations]. nodes represent proteins and edges represent physical interactions; in a ‘transcriptional regulatory network’, nodes represent transcription factors or DNA regulatory ele- Omics and Personalized Medicine for IBD ments and edges represent physical bindings; in a ‘meta- bolic network’, nodes represent enzymes and edges repre- Some of the very traditional tools used to diagnose and sent metabolites, and in a ‘disease network’, nodes rep- monitor patients, such an endoscopic mucosal biopsy, resent diseases and edges represent genetic variants can be used for far more sophisticated purposes and as a

IBD Omics Dig Dis 2014;32(suppl 1):96–102 99 DOI: 10.1159/000367836 Downloaded by: C.H.U. de Vigo 198.143.45.65 - 4/7/2016 12:11:31 PM source of omics data for tailoring therapy [15] . Going be- prednisolone is already known to be effective in IBD, yond its mundane use for histological and immunohisto- topiramate, which is presently used for a distinct clinical chemical analyses, a routine mucosal biopsy is a precious indication, is not. Nevertheless, when topiramate was giv- biospecimen that can be submitted to molecular and en to animals with experimental colitis, it was highly ef- pathway analysis as well as genetic analysis for subse- fective in reducing inflammatory scores. This shows the quent pharmacogenomics, paving the way to personal- value of these new bioinformatics-based tools and affords ized medicine. These advanced bioinformatics-based an- proof of principle of their potential for unsuspected new alytical methodologies form the basis to implement what drug discovery. has been termed an ‘integrative personalized omics pro- file’ (iPOP) [16] . To obtain an individualized single patient-derived iPOP, a biospecimen can be submitted to Omics and the Future of IBD DNA and RNA extraction or microbiota and metabolite profiling, for example, and yield information on the ge- Is there still a role for the physician in charge of diag- nome, epigenome, metagenome, transcriptome, pro- nosing and managing patients with IBD in this rapidly teome and metabolome of the patient. This creates an ini- approaching world of omics? Or, perhaps more correct- tial iPOP pattern that is then repeated prospectively over ly, what will his or her new role be? In some ways, accord- time during the course of the disease, generating a series ing to the definition of omes and omics, what the practic- of iPOPs that reflect the disease as a whole at different ing gastroenterologist does in daily practice is a sort of time points and thus the complete disease evolution. This ‘clinical omics’ ( fig. 3 ). Obtaining a medical and family information is eventually fed back to the managing physi- history; collecting signs and symptoms; performing cian for custom-made pharmacogenomics-based treat- physical examinations; securing information by ordering ment [16]. This highly integrated and sequential ap- blood chemistry, serology, cultures and stool examina- proach will also facilitate the identification of patients tions; performing a series of imaging studies, and doing with the same or very similar underlying set of pathogen- endoscopies with biopsies are all in reality the gathering ic events and thus identify fairly homogeneous popula- of multiple data to be utilized for diagnostic and manage- tions of patients that will benefit from customized treatment purposes. So, in this regard, practicing routine ment instead of a random, same-for-all treatment for medicine can still be seen as a form of omics, but at a subjects with the same clinical diagnosis [17] . So, instead sensibly more rudimentary level compared to what using of having a 30–40% beneficial response to treatment in a bioinformatics tools can do. The current basic methods random and heterogeneous population of IBD patients, of clinical diagnosis and treatment are obviously justified personalized medicine would provide a much higher suc- and still indispensable, but even today clinical data can cess rate in an omics-characterized homogeneous sub- be blended with molecular and genetic data and integrat- population of patients. ed for the formation of network models. A report by Xing Not only can omics approaches help by selecting the et al. [20] shows how a disease model can be built by com- best medication for the right patient, but they can also ac- bining genetics, whole blood gene expression profiles celerate drug discovery, and reduce the current 12-year and clinical parameters of rheumatoid arthritis patients average period for new drug discovery to less than 4 years to identify novel therapeutic interventions for patients [18] . An example of how this can be accomplished has that fail to respond to TNF blockers. Sooner than we an- recently been reported by Dudley et al. [19] with the tech- ticipate, clinical omics will be progressively replaced by nique of computational repositioning for new drug dis- far more sophisticated methodologies that will yield covery for IBD. Using this technology, mRNA signatures more abundant and more detailed information at an un- that increase or decrease in a particular human disease are precedented pace and depth. This will inevitably be uti- compared to mRNA expression signatures from human lized for therapeutic purposes. In fact, the National Can- cells exposed to drugs; then a similarity score is calculated cer Institute of the United States has already developed ranging from a perfect correlation to an opposite correla- specific criteria for the use of omics-based predictors in tion, and the presumption is made that an opposite cor- clinical trials in which omics should be used to guide relation would counteract the disease abnormalities and therapy [6] . Analogous criteria will be eventually devel- be therapeutically effective. Computational repositioning oped for any disease, including IBD, opening the door to identified 2 top compounds effective in IBD, which were personalized medicine, more specific therapies and im- prednisolone and the anticonvulsant topiramate. While proved outcomes.

100 Dig Dis 2014;32(suppl 1):96–102 Fiocchi DOI: 10.1159/000367836

Downloaded by: C.H.U. de Vigo 198.143.45.65 - 4/7/2016 12:11:31 PM Blood chemistry

Physical examinationon Serology

Symptoms Cultures Color version available online

Family Stool history examination Clinical History X rays IBD ‘omics’

Immuno- histochemistry CT scan

Histology MRI

Fig. 3. Clinical omics in IBD. By perform- Biopsies Ultrasound ing routine diagnostic evaluations, the per- son managing an IBD patient gathers a se- Endoscopy ries of disparate data that are subsequently put together for diagnostic and therapeutic purposes.

IBD ‘omics’: how can I study them?

Bioinformatics experts Edges Color version available online Fig. 4. Complementary knowledge needed Nodes to implement omics-based systems biology studies in IBD. IBD investigators can generate patient biospecimen-derived infor- Bioinformatic mation that defines multiple omes forming tools the nodes of the biological interactions underlying IBD pathogenesis. Bioinformatics experts analyze the omes and determine the number and direction of the edges IBD IBD Systems among them to identify the hubs or central investigators biology regulators of IBD pathogenesis. Once these data are integrated, additional bioinformatic tools generate IBD-specific systems biology-based profiles that can be used for IBD ‘omics’: what you want to know? diagnostic and therapeutic purposes.

Access to this expanding world of omics is already here interactome.dfci.harvard.edu). IBD-specific websites are and a number of Web-based resources are available to also available, such as IBDsite (http://www.itb.cnr.it/ibd) investigators eager to define the genome, exome, epi- [21]. Obviously the creation and use of these omics-based genome, methylome, regulome, transcriptome, editome, tools require very specific expertise, one that it is and like- miRNome, proteome, autoantibodyome, metabolome, ly will continue to be out of reach of basic and clinical IBD microbiome and pharmacogenome of their patients [16] . investigators interested in using omics-derived informa- Even an interactome website is ready for use where net- tion for clinical purposes. Fortunately, new generations works for all omics interactions can be probed (http:// of mathematically oriented scientists with a focus on hu-

IBD Omics Dig Dis 2014;32(suppl 1):96–102 101 DOI: 10.1159/000367836 Downloaded by: C.H.U. de Vigo 198.143.45.65 - 4/7/2016 12:11:31 PM man biology have emerged and are quickly expanding actions (the edges) and regulatory pathways among them, their ranks, and they will be ones ‘doing the math’ for and then feed this information back to the clinical side of disease omics goals [22] . Several systems networks al- IBD. ready exist that can do so. An example are Bayesian net- If we expect, hope for or want better ways to treat pa- works that have the capacity to unravel interactions tients with IBD, we have to openly accept the complexity among disease, environment and genes [23] , and have of IBD and whole new ways to look at it, explore it and been used to analyze gene expression, signaling pathways, learn from it, including all of the associated omes and protein-protein interactions, genetic epidemiology and omics. Each IBD patient has his or her own IBD inte- disease recurrence [24]. Nevertheless, to be able to take grome that is unique and composed of innumerous omes full advantage of the power of these models for the ben- all mutually interacting among themselves through an efit of IBD research and ultimately IBD sufferers, it will ‘IBD interactome’. This realization may be intimidating be essential to establish close and robust complementary or overwhelming, but it must be accepted and dealt with ties and expertise between IBD investigators and bioin- if we want to make progress in the challenging and fasci- formatics experts who will have to work together, albeit nating field of IBD. in distinct specific roles (fig. 4 ). The IBD investigators will be the ones who, by their ability to access patients and se- cure their biospecimens, can generate data to create rel- Disclosure Statement evant omes (the nodes); the bioinformatics experts will analyze these data to study and identify the various inter- None.

References

1 Xavier RJ, Podolsky D: Unraveling the patho- 9 Goh KI, Choi IG: Exploring the human dis- 17 Chan AC, Behrens TW: Personalizing medicine genesis of inflammatory bowel disease. Nature easome: the human disease network. Brief for autoimmune and inflammatory diseases.

2007; 448: 427–434. Funct Genomics 2012; 11: 533–542. Nat Immunol 2013; 14: 106–109. 2 Renz H, von Mutius E, Brandtzaeg P, Cookson 10 Dimitrov DV: The human gutome: nutrige- 18 Kraljevic S, Stambrook PJ, Pavelic K: Accelerat-

WO, Autenrieth IB, Haller D: Gene-environ- nomics of the host-microbiome interactions. ing drug discovery. EMBO Rep 2004; 5: 837–842.

ment interactions in chronic inflammatory dis- OMICS 2011; 15: 419–430. 19 Dudley JT, Sirota M, Shenoy M, Pai RK, Roed-

ease. Nat Immunol 2011; 12: 273–277. 11 Arijs I, Li K, Toedter G, Quintens R, Van Lom- der S, Chiang AP, et al: Computational reposi- 3 Fischer HP: Mathematical modeling of complex mel L, Van Steen K, et al: Mucosal gene signa- tioning of the anticonvulsant topiramate for in-

biological systems: from parts lists to under- tures to predict response to infliximab in pa- flammatory bowel disease. Sci Transl Med 2011;

standing systems behavior. Alcohol Res Health tients with ulcerative colitis. Gut 2009; 58: 1612– 3: 96ra76.

2008; 31: 49–59. 1619. 20 Xing H, McDonagh PD, Bienkowska J, Cashora- 4 Morris T, Stables M, Colville-Nash P, Newson J, 12 Planell N, Lozano JJ, Mora-Buch R, Masamunt li T, Runge K, Miller RE, et al: Causal modeling Bellingan G, de Souza PM, et al: Dichotomy in MC, Jimeno M, Ordas I, et al: Transcriptional using network ensemble simulations of genetic duration and severity of acute inflammatory re- analysis of the intestinal mucosa of patients with and gene expression data predicts genes in- sponses in humans arising from differentially ulcerative colitis in remission reveals lasting ep- volved in rheumatoid arthritis. PLoS Comput

expressed proresolution pathways. Proc Natl ithelial cell alterations. Gut 2013; 62: 967–976. Biol 2011; 7:e1001105.

Acad Sci USA 2010; 107: 8842–8847. 13 Leal RF, Planell N, Kajekar R, Lozano JJ, Ordas 21 Merelli I, Viti F, Milanesi L: IBDsite: a Galaxy- 5 Jostins L, Ripke S, Weersma RK, Duerr RH, Mc- I, Dotti I, et al: Identification of inflammatory interacting, integrative database for supporting Govern DP, Hui KY, et al: Host-microbe inter- mediators in patients with Crohn’s disease un- inflammatory bowel disease high throughput

actions have shaped the genetic architecture of responsive to anti-TNFα therapy. Gut 2014, data analysis. BMC Bioinformatics 2012;

inflammatory bowel disease. Nature 2012; 491: Epub ahead of print. 13(suppl 14):S5. 119–124. 14 De Preter V, Machiels K, Joossens M, Arijs I, 22 Rezola A, Pey J, Tobalina L, Rubio A, Beasley JE, 6 McShane LM, Cavenagh MM, Lively TG, Eber- Matthys C, Vermeire S, et al: Faecal metabolite Planes FJ: Advances in network-based metabol- hard DA, Bigbee WL, Williams PM, et al: Crite- profiling identifies medium-chain fatty acids as ic pathway analysis and gene expression data in- ria for the use of omics-based predictors in clin- discriminating compounds in IBD. Gut 2014, tegration. Brief Bioinform 2014, Epub ahead of

ical trials. Nature 2013; 502: 317–320. Epub ahead of print. print. 7 Polytarchou C, Koukos G, Iliopoulos D: Sys- 15 Basik M, Aguilar-Mahecha A, Rousseau C, Diaz 23 Su C, Andrew A, Karagas MR, Borsuk ME: Us- tems biology in inflammatory bowel diseases: Z, Tejpar S, Spatz A, et al: Biopsies: next-gener- ing Bayesian networks to discover relations be- ready for prime time. Curr Opin Gastroenterol ation biospecimens for tailoring therapy. Nat tween genes, environment, and disease. BioDa-

2014; 30: 339–346. Rev Clin Oncol 2013; 10: 437–450. ta Min 2013; 6: 6. 8 Vidal M, Cusick ME, Barabasi AL: Interactome 16 Li-Pook-Than J, Snyder M: iPOP goes the 24 Rodin AS, Boerwinkle E: Mining genetic epide-

networks and human disease. Cell 2011; 144: world: integrated personalized Omics profiling miology data with Bayesian networks I: Bayesian 986–998. and the road toward improved health care. networks and example application (plasma apoE

Chem Biol 2013; 20: 660–666. levels). Bioinformatics 2005; 21: 3273–3278.

102 Dig Dis 2014;32(suppl 1):96–102 Fiocchi DOI: 10.1159/000367836

Downloaded by: C.H.U. de Vigo 198.143.45.65 - 4/7/2016 12:11:31 PM