Genome-wide association studies (GWAS) have provided pivotal insights into understanding complex trait genetic architecture and its evolutionary processes. Findings from GWAS suggested that negative selection has shaped the genetic architecture by suppressing alleles with large trait-effect sizes at a low frequency in the population. According to classic theory (Maruyama 1974), these alleles are expected to be younger than neutral alleles. However, not until recently, empirical estimates of allelic age have become available for genome-wide variants. This study aims to combine these estimates with GWAS data to draw evolutionary inference on the trait-associated variants across functional genomic categories.
We analysed data from the Atlas of Variant Age and genetic fine-mapping results for 94 complex traits in the UK Biobank. We found that variants with nonzero effects on at least one trait (minimal posterior inclusion probability or PIP >= 0.9) mostly arose from 7,629 generations ago (median age), co-occurred with the origin of modern humans in Africa (before the out-of-Africa bottleneck). Next, we randomly sample control variants, matching their minor allele frequencies and linkage disequilibrium scores with those of trait-associated variants. Result showed that the fine-mapped variants were significantly younger than control variants, with their differences in allelic age increased with the PIP threshold. This result is consistent with the theoretical prediction under the model of negative selection on the trait-associated variants.
Furthermore, partitioning the trait-associated variants into functional genomic categories revealed variable allelic age distributions across categories, with coding regions enriched most in recent variants (median age of 1,317 generations versus 6,039 generations for control variants sampled from the same category). Across functional categories, the median allelic age was positively correlated (r = 0.39) with the estimate of S parameter (Zeng et al 2021), suggesting a convergence of different signatures of natural selection in the functional genetic architecture of complex traits.