Genome-wide association studies of type 2 diabetes (T2D) show common variant heritability is enriched within gene regulatory regions of pancreatic islets. Whole genome sequence (WGS) data in cohorts allows us to determine the contribution of the complete spectrum of genetic variation to the heritability of T2D risk, including rare alleles that have recently arisen in the human population. A variant-based heritability analysis can partition disease risk into environmental and genetic components, and the genetic components can be further decomposed by variant frequency (a proxy of variant age) and pancreatic islets regulatory annotation (a proxy of functional relevance to T2D risk). We hypothesized that rare, non-protein coding variants would have a significant contribution to the heritability of T2D risk and that further restricting to gene regulatory regions active in pancreatic islets would also produce significant heritability estimates. We conducted a WGS heritability analysis (GREML method in GCTA software) of T2D risk in 15109 unrelated individuals (2215 with T2D; 12894 controls) of European ancestry from the NHLBI’s TOPMed study. We partitioned variants with allele count > 5 by minor allele frequency (MAF): rare variants [MAF<0.1%], and annotation: all non-coding variants, and non-coding variants in pancreatic islets regulatory regions. We further partitioned variants by linkage disequilibrium properties to reduce estimate bias. One of the rare, "all non-coding" variants sets showed a relatively large liability-scale heritability estimate (h2=0.6, P< 0.001). Further significant estimates in rare variants were from active transcription start sites (h2=0.1, P<0.01); active promoters (three variant sets with h2=0.07, 0.08 and 0.1, P<0.05, after body mass index adjustment) and active enhancers from promotor capture Hi-C models (h2=0.07, P<0.05). These findings show a role for rare, non-coding variants in the genetic architecture of type 2 diabetes.
A. Manning: None.
National Institutes of Health (K01DK107836)