been performed in populations of Caucasian ancestry, however the overlap of GWAS SNPs to the strongest-associated cis-eQTLs did not reflect this; instead all populations were represented (CEU: N = 14 SNPs, CHB: N = 10, GIH: N = 9, JPT: N = 21, LWK: N = 8, MEX: N = 10, MKK: N = 9, YRI: N = 8). Of the 62 SNPs that were the most significant cis-eQTL for a given gene as well as associated to a trait in the GWAS catalog, we observed that 15 (∼24%) were the most significant SNP of the same gene in at least one additional population, a proportion which is likely an underestimate of the true value, given that the same SNPs weren't necessarily tested in all populations. We asked whether across-population replication of the most significant SNP per gene differed depending on whether that SNP was a GWAS SNP, and observed no difference (Fisher's 2-tailed p-value 0.128). The eQTLs identified through our analyses contribute significantly to the available functional regulatory data that could be used to fine-map and elucidate the function of genetic variants contributing to complex phenotypes.
