Authors: Hakhamanesh Mostafavi; Jeffrey Spence; Sahin Naqvi; Jonathan Pritchard
Affiliation: Stanford University
Keywords: Complex traits
Most findings in genome-wide association studies (GWAS) of complex traits point to non-coding genetic variants with putative gene regulatory effects. However, currently identified expression quantitative trait loci (eQLTs) explain only a small fraction of the GWAS signals. While this lack of overlap may at first appear to be counterintuitive, we wondered if this might be a natural consequence of population-genetic forces such as natural selection. By analyzing GWAS hits for complex traits in the UK Biobank and cis-eQTLs from the GTEx consortium, we show that these assays systematically discover different types of genes and variants: eQTLs cluster strongly near transcription start sites, while GWAS hits do not. Genes near GWAS hits are enriched in numerous functional annotations, are under strong selective constraint and have a complex regulatory landscape across different tissue/cell types, while genes near eQTLs are depleted of most functional annotations, show relaxed selection, and have a simpler regulatory landscape. We describe a model to understand these observations. Specifically, modeling selection on complex traits, we demonstrate that selection disproportionally hampers the discovery of functionally relevant eQTLs, particularly eQTLs regulating genes that contribute most to trait heritability. More broadly, our work provides insight into the utility of intermediate phenotypes for explaining genetic effects on complex traits.