To empirically examine the concordance of genetic effects on T2D across different populations, we extracted the lead variant of each genome-wide significant locus in the fixed-effect meta-analysis of the three GWAS and compared their per-allele effect sizes across populations. When the lead variant was missing in a population, we used the tag variant that was most strongly correlated with the lead variant with an R2 no smaller than 0.6. We note that the MEDIA T2D GWAS was imputed to HapMap reference panels and included a relatively small number of genetic variants (~2.5M) across the genome. As a result, the large majority of the variants representing the African population were tag variants which may lead to under-estimation of the effect size concordance between ancestries.
