241-OR: Causal Gene Candidates for Type 2 Diabetes Based on Protein-Coding Variants in 127,676 Individuals

Genome-wide association studies (GWAS) have identified >250 genetic loci associated with type 2 diabetes (T2D), but, currently, few associations have been translated to causal genes. To directly implicate genes in T2D pathogenesis, we analyzed coding variants from the largest whole exome sequencing (WES) study to date, consisting of 36,652 cases and 91,024 controls from the Accelerating Medicines Partnership-T2D-GENES, Cohorts for Heart and Aging Research in Genomic Epidemiology, Myocardial Infarction Genetics, and NHLBI Trans-Omics for Precision Medicine consortia. Through single-variant and gene-level meta-analysis, we identify three evidence tiers of T2D-susceptibility genes. First, 11 genes harbor exome-wide significant missense associations (p≤4.3x10^-7) and four genes have significant gene-level associations (p≤2.5x10^-6). Two of these associations are novel: a missense mutation in MET (p=2.33x10^-7), a gene with no previous genetic link to T2D, and, while presence of misdiagnosed MODY cases cannot be excluded, a series of rare alleles in GCK providing the first exome-wide significant association with common T2D. Second, we apply a gene-level Bayesian statistical approach to identify 29 genes with posterior probabilities (PPA) of ≥50% of explaining signals at GWAS loci. Twenty-one of these genes lie within loci where no causal gene is known, including SGP7 (PPA=0.85), a zinc metalloprotease that contains a previously reported independent common missense variant associated with T2D. Finally, we extend this Bayesian approach to another 1,069 genes previously studied in mice. This analysis identifies 17 genes with gene-level association PPA ≥50%, including 11 genes with no prior human genetic link to T2D such as ONECUT1 (PPA=0.54) which is exclusively expressed in liver and pancreas. Our results further illustrate the value of WES in identifying genes that may be future T2D drug targets. All results presented will be available through the AMP-T2D Knowledge Portal.