An examination of previous datasets confirms a general trend to consistency of GWAS results across continents and emphasizes the benefits of incorporating as many associations as possible. Fu et al. [21] focused on associated SNPs discovered in East Asian GWAS. Although they used only four traits and 47 SNPs (43 loci), they demonstrated the challenges of multi-ethnic studies, and provided a framework to cope with these difficulties. As discussed by the authors, caution is warranted as to whether the disease loci and/or causal variants are population-specific. For instance, they suggested that two signals for type 2 diabetes located in 9p24.1 (PTPRD, rs17584499) and 17p13.3 (SRR, rs391300) could be East Asian-specific, as they fail to replicate in a well-powered study in Europeans. However, we gathered several replication attempts of these two signals in more recent East Asian GWAS (Table S4), and out of 8 replication attempts only one was successful (for PTPRD, rs17584499) at P<0.05, when a total of 7.49 replications would be expected by power alone (4 for PTPRD and 3.49 for SRR, see Table S5). Also, in only 4
