Accurate and Rapid Genome Interpretation – in Clinical Care
Total Page:16
File Type:pdf, Size:1020Kb
Accurate and Rapid Genome Interpretation – in clinical care Martin Reese, Ph.D. Overview § Who we are? § Our unique bioinformatics technologies VAAST/Phevor - ready to scale § Three clinical applications - Discovery novel disease genes and diagnoses - The Utah Genome Project - Diagnostic Odyssey patients – Genomics England 100,000 Genome Project - NICU/ICU babies – Stephen Kingsmore’s Rady’s Children Hospital Fabric GenomicsTM 2 Company Overview § Company - Founded 2009 as Omicia § Product - Comprehensive end-to-end Precision Medicine platform for genomic data analysis (SaaS) - OpalTM Clinical is a single, scalable unified clinical database solution (>65,000 genomes) § Strategic Collaborators - University of Utah: VAASTPHEVOR papers published in 2011 by Yandell, Reese et al., originally funded by multiple NIH grants § Customer Segments - Clinical Labs, Hospital Labs, Pharmaceutical Research and Country Projects - >1,000 research and clinical licensees, >25 clinical labs supported Fabric Genomics, Inc. 3 The Genomic “A high quality clinical genome is now <$2,000 but it takes 200 Challenge Now person hours for interpretation, is Analyzing and which is more than $20,000” Michael Snyder, Director Genomics & Interpreting Personalized Medicine, Stanford all of the Data Fabric Genomics, Inc. v Variant Interpretation § How do we find the variants important for Hereditary Breast Cancer: Extended HereditaryPowered Breast by Fabric Cancer:TM Enterprise Extended Powered by Omicia's Opal Clinical Interpretation Platform PATIENT: Maria Fox SPECIMEN TYPE: Blood ORDERING PHYSICIAN: Dr. Mary Worth DOB: 6/4/1963 COLLECTED: 01/04/2016 REPORT DATE: Jan 21, 2016 the disease or phenotype of the ACCESSION ID: DEMO1982 RECEIVED: 01/05/2016 SEX: Female ETHNICITY: Hispanic patient(s)? INDICATION FOR TESTING V16.3 - Family history of malignant neoplasm, breast Test Result: Positive. A pathogenic variant was identified in BRCA1. PRIMARY FINDINGS Gene Variant Position Classification Zygosity BRCA1 c.4097-1G>A g.41243050C>T pathogenic heterozygous § How do we find the important variants PRIMARY FINDINGS SUMMARY This individual is reported to have a family history of breast cancer, with the paternal grandmother and aunt affected. This patient is heterozygous for variant c.4097-1G>A associated with the BRCA1 gene. This interpretation identified a variant in BRCA1 that has been previously classified as pathogenic (PMID: 17924331). This mutation is in a splice acceptor region, and a previous case was associated with breast cancer in a single quickly, reliably, and reproducibly within family, co-segregating with disease. Inherited mutations of BRCA1 are responsible for about 40-45% of hereditary breast cancers; however these mutations account for only 2-3% of all breast cancers, since the BRCA1 gene is rarely mutated in sporadic breast cancers. Most of these mutations lead to the production of an abnormally short version of the BRCA1 protein, or prevent any protein from being made from one copy of the gene. a team of variant scientists? Other BRCA1 mutations change single protein building blocks (amino acids) in the protein or delete large segments of DNA from the BRCA1 gene. Researchers believe that a defective or missing BRCA1 protein is unable to help repair damaged DNA or fix mutations that occur in other genes. As these defects accumulate, they can allow cells to grow and divide uncontrollably and form a tumor. RECOMMENDATIONS Genetic counseling is recommended to discuss the implications of this finding for the patient. Genetic counseling and testing for at risk family members is also recommended. For more information or questions please contact customer services at 415.123.4567. TEST METHODOLOGY Omicia's extended Hereditary Breast Cancer panel includes validated high penetrance and moderate penetrance breast cancer susceptibility genes recognized by the US National Cancer Institute. These are mostly § How do we find the important variants part of inherited cancer syndromes and include the highly penetrant autosomal dominant genes BRCA1 and BRCA2 (Hereditary Breast/Ovarian Cancer Syndrome), PTEN (Cowden Syndrome), CDH1 (Hereditary Diffuse Gastric Cancer), TP53 (Li Fraumeni Syndrome) and STK11 (Peutz-Jeghers Syndrome) and the moderately penetrant autosomal dominant genes ATM (Ataxia telangiectasia), PALB2 (Fanconi anemia) and CHEK2. TruSight One kit (v1.0) was used to target the exon regions of genes. These targeted regions were sequenced at scale? using the NextSeq 500 sequencing system with 150bp paired-end reads. Sequences were aligned to the human Patient: Maria Fox Accession: DEMO1982 DOB: 6/4/1963 Age/Sex: 52 / Female Report Date: Jan 21, 2016 Page 1 of 2 Fabric GenomicsTM 5 is now NEXT-GEN SECONDARY INTERPRETATION CLINICAL SEQUENCING ANALYSIS AND REPORTING MANAGEMENT Fabric Genomics’ End to End Platform Variant Annotation Gene Discovery Clinical Systems Alignment Calling & Analysis & Interpretation Reporting Integration Fabric GenomicsTM VAAST and Phevor 6 • VAAST: probabilistic disease gene finder • Phevor: belief network for integration of phenotype data via HPO and GO with genome sequences A probabilistic disease-gene finder for personal genomes Yandell M, Huff C, Hu H, Singleton M, Moore B, Xing J, Jorde LB, Reese MG: Genome Res 2011, 21(9):1529-1542. pVAAST: unified means for linkage analysis and rare-variant association. Hu H., Roach JC, Coon H, Guthery SL, Tavtigian S, Shankaracharya D., Wu W, Xing J, Glusman G, Hubley R, Li H, Garg V, Moore M, Holloway A, Hood L, Pollard KS, Galas DJ, Srivastava D, Reese MG, Jorde LB, Yandell M, Huff CD (2014) Nature Biotechnology, 32(7), 663–669. Phevor Combines Multiple Biomedical Ontologies for Accurate Identification of Disease-Causing Alleles in Single Individuals and Small Nuclear Families. Singleton M., Guthery SL., Voelkerding KV., Chen K., Kennedy BJ., Margraf RL., Durtschi J., Eilbeck K., Reese MG., Jorde LB., Huff CD., Yandell M. Am J Hum Genet. 2014 Apr 3;94(4):599-610. A probabilistic disease-gene finder for personal genomes Yandell M, Huff C, Hu H, Singleton M, Moore B, Xing J, Jorde LB, Reese MG: Genome Res 2011, 21(9):1529-1542. VAAST 2.0: Improved Variant Classification and Disease-Gene Identification Using a Conservation-Controlled Amino Acid Substitution Matrix. Hu, H., C.D. Huff, B. Moore, S. Flygare, M.G. Reese, and M. Yandell, Genet Epidemiol, 2013. 37(6): p. 622-34. pVAAST: unified means for linkage analysis and rare-variant association. Hu H., Roach JC, Coon H, Guthery SL, Tavtigian S, Shankaracharya D., Wu W, Xing J, Glusman G, Hubley R, Li H, Garg V, Moore M, Holloway A, Hood L, Pollard KS, Galas DJ, Srivastava D, Reese MG, Jorde LB, Yandell M, Huff CD (2014) Nature Biotechnology, 32(7), 663–669. Using VAAST to Identify Disease-Associated Variants in Next-Generation Sequencing Data. Brett Kennedy, Zev Kronenberg, Hao Hu, Barry Moore, Steven Flygare, Mark Yandell, Chad Huff. Current Protocols 24 April 2014. VAAST rhymed with BLAST BLAST VAAST query case genomes database control genomes (1K Genomes Project) hits hits Expect P-value Fast Fast BLAST searches for statistically significant similarity between sequences. VAAST searches for statistically significant dissimilarity between sequences. It uses a Burden Test to do so. VAAST work supported by NIH SBIR grants 1R4HG003667 to Fabric/Yandell, SBIR 1R44HG002991 to Fabric, an NIH ARRA GO grant 1RC2HG005619-01 to Yandell/Reese, all administered by the National Human Genome Research Institute (NHGRI). VAAST+ by R01GM104390 from the NIGMS to Yandell and NSF DEB-1149160 to M. Shapiro Why is a Burden calculation necessary? • Variant prioritization only gets you partway there. • Consider an individual with a de novo highly damaging missense allele, e.g. His-> Trp. • Judged to be maximally damaging by Sift, PolyPhen etc. • Allele is completely absent from gnomAD and 1KG. • Is this a allele responsible for the proband’s genetic disease? • But what if 20% of gnomAD and 1KG individuals carry one or more different, but equally ‘damaging’ alleles in the same gene? Why is a Burden calculation necessary? • The VAAST p-value speaks to this question, i.e. what % of the population have greater than the observed genotype burden at that locus. • Every variant in the population, every genotype is considered in this calculation. • Additional benefits include that it controls for LD, ethnic stratification, and unknown sources of noise in the data, etc. • VAAST also models phenomena such as mode of inheritance, penetrance, disease prevalence, etc. VAAST, Families and Disease Recessive inhr model Complete penetrance A family with a Tay-Sachs affected child Slide 40 Phenotype Driven Variant Ontological Re-Ranking Tool Phevor Combines Multiple Biomedical Ontologies for Accurate Identification of Disease-Causing Alleles in Single Individuals and Small Nuclear Families. Singleton M., Guthery SL., Voelkerding KV., Chen K., Kennedy BJ., Margraf RL., Durtschi J., Eilbeck K., Reese MG., Jorde LB., Huff CD., Yandell M. Am J Hum Genet. 2014 Apr 3;94(4):599-610. Patient phenotype (symptoms) are big clues Phevor Combines Phenotype and Genotype Phenotype description Improved power to identify causative variants Phevor Combines Multiple Biomedical Ontologies for Accurate Identification of Disease-Causing Alleles in Single Individuals and Small Nuclear Families. Singleton M., Guthery SL., Voelkerding KV., et al., Reese MG., Jorde LB., Huff CD., Yandell M. ( 2014) . Am J Hum Genet. 2014 Apr 3; 94(4):599-610. Fabric GenomicsTM 15 What problem does Phevor solve? Search