Inter-population differences of allele frequency and regulome tagging are associated with the heterogeneity of loci identified by cross-ancestry genome-wide association studies

To provide novel insight regarding the inter-population diversity of loci associated with complex traits, we integrated genome-wide data from UK Biobank (UKB) and 1,000 Genomes Project (1KG) data representative of the genetic diversity among worldwide populations. We investigated genome-wide data of 4,359 traits from 361,194 UKB participants of European descent. Using 1KG data, we explored the allele frequency differences and linkage disequilibrium (LD) structure of UKB genome-wide significant (GWS) loci across worldwide populations. Functional annotation data were used to identify regulatory elements and evaluate the tagging properties of GWS variants. No significant difference was observed in allele frequency between UKB and 1KG GBR (British in England and Scotland). Considering other population groups, we identified genome-wide significant alleles with frequencies different from what expected by chance: UKB vs. 1KG Europeans without GBR (rs74945666; allele=T [0.908 vs. 0.03], standing height pGWAS=1.48x10-17), UKB vs. 1KG African (rs556562; allele=A [0.942 vs. 0.083], platelet count pGWAS=4.84x10-15), UKB vs. 1KG Admixed Americans (rs1812378; allele=T [0.931 vs. 0.089], standing height pGWAS=4.23x10-12), UKB vs. 1KG East Asian (rs55881864; allele=T [0.911 vs. 0.001], monocyte count pGWAS=7.29x10-13), and UKB vs. South Asian (rs74945666; allele=T [0.908 vs. 0.061], standing height pGWAS=1.48x10-17). LD-structure analysis and computational prediction showed differences in how these alleles tag functional elements across human populations. In conclusion, the human diversity of certain GWS loci appear to be affected by local adaptation while in other cases the associations may be biased by residual population stratification.