TOPMed Whole Genome Sequence Combined with Pancreas-Specific Annotation for Rare Variants Tests of Type 2 Diabetes Risk

Genetic studies have shown that common variants associated with type 2 diabetes are enriched in pancreatic-specific regulatory regions. Most whole genome sequence (WGS) studies have used general annotation of regulatory regions not specific to relevant tissues. In contrast, we used functional annotation from pancreatic islets to define rare variant groupings for aggregate tests in coding and non-coding regions in a pilot analysis on chr 10 with 13,200 individuals across six NHLBI TOPMed studies: 2,601 individuals with type 2 diabetes and 10,599 normoglycemic controls with <30X WGS coverage. We identified genes expressed in pancreas and nearby regions showing islet-specific regulatory activity, including open chromatin and chromatin states associated with promoters and enhancers. Variants with minor allele frequency < 1% were aggregated by gene, including the region 5 kbp upstream and downstream of transcribed regions. We only included variants annotated as: "splice acceptor", "splice donor", "splice region", "missense", or "stop gained" within exons expressed in islets, while outside of those exons we included only those variants within “active transcription start sites” and “active enhancer” predicted chromatin states. We performed SKAT association tests, adjusting the association with type 2 diabetes for age, sex, study, and a genetic relatedness matrix to accounted for genetic similarities within families and between populations. Overall, 219 genes (median of 161 variants per gene-region) expressed in islets were included in the analysis: ATP5C1, AVPI1, and YME1L1 reached nominal significance (P<0.01). Chr 10 pilot analyses provide a framework for tailoring rare variant analyses in WGS data through the use of disease-specific functional annotation. WGS analysis in larger samples and additional functional data generated from collaborators in the NIDDK’s AMP-T2D Consortium will extend the genomic spectrum associated with type 2 diabetes risk.