Integrative computational-experimental approaches to stratify monogenic disease risk Project Summary/Abstract Despite the availability of large disease cohorts and national biobanks, it remains challenging to interpret the functional and clinical impact of rare missense variants in established disease genes. Even in genes like BRCA1 and LDLR, most nonsynonymous variants in these genes are rare, and so there are often too few individuals to estimate risk epidemiologically. Functional screening assays have shown great promise toward resolving the impact of these missense variants, but to date, they have been too costly to scale to many genes and phenotypes. This dearth of human genetics evidence and absence of functional data has resulted in many ‘Variants of Uncertain Significance’ (VUS), which poses challenges in clinical management for patients. However, we are now at a turning point on two fronts: 1) biobanks have sequenced the exomes of hundreds of thousands of individuals (e.g. UK Biobank, Geisinger MyCode, All of Us) and 2) cost effective variant installation assays enable the functional assessment of thousands of coding variants at a time. These two approaches have complementary benefits and drawbacks. Rare variant burden analyses from exomes leverage clinical information, yet for the vast majority of individual coding variants there are insufficient numbers of carriers to provide confident estimates of risk. Functional assays provide robust data on individual variants, yet in vitro phenotypes may be imperfect surrogates for clinical phenotypes. In this proposal, we combine both techniques, developing a computational method that integrates clinical phenotyping and in vitro functional assessment to improve coding variant risk assessment. We will use this approach to estimate the risk for all variants in 100 genes associated with cancer and CAD. This project will provide high quality functional and human genetics evidence for a broad set of phenotypes and genes, and we will measure its impact and ability to scale genomic medicine by reassessing VUSs in two large biobanks. We selected 100 genes with broad clinical impact, with an established functional assay, and existing support for disease association from biobank cohorts. First, we develop an optimized variant installation library that covers all known rare variants from over one million sequenced individuals, and variants previously been assessed in the diagnostic setting (ClinVar). Second, we use related assay measurements to improve the quality of screening estimates, and infer every possible substitution within these genes. This strategy will help groups scale their studies to cover many more variants using rational library design. Finally, we integrate these epidemiological and functional data to characterize the clinical risk of variants in at least 100 genes associated with cancer and CAD. In at least 38 genes recommended for secondary findings analysis by the American College of Medical Genetics or ClinGen, we will apply evidence in the form of functional impact and case data to update VUS classifications in the Geisinger MyCode and Mass General Brigham biobanks, representing over 250,000 individuals, using established disclosure processes, based on patient consent and preferences.

Integrative computational-experimental approaches to stratify monogenic disease risk Project Narrative The interpretation of variants in monogenic disease genes (e.g. ACMG Secondary Findings list) remains challenging, despite the availability of large cohort and population health studies. Even in well-known disease genes, the majority of mutations are only observed in one or two families, which makes it difficult to understand their clinical risk using epidemiological data. This project combines experimental screening and population data in 100 genes with known monogenic disease risk to dramatically improve risk assessments for these variants.