All these inferences are confirmed after categorizing European discovery GWAS into two groups using a threshold of 10,000 individuals to distinguish between “small” and “large” studies. First, we did not observe differences in the MAF distribution of associated SNPs according to the discovery sample size (average MAF of 0.301 vs. 0.333 for “small” and “large” GWAS respectively; P = 0.12, Wilcoxon test). Second, even if larger GWAS do indeed detect associations with smaller ORs (average OR 1.15 vs. 1.28; P<3×10−7), the trans-continental correlation of ORs between Europeans and East Asians was the same for “small” and “large” GWAS (Figure 7). Both results show, yet again, that causal variants of different effect sizes are equally shared across populations, independently of the sample size of the discovery GWAS.
