As mentioned above, we hypothesized that allele ΔFs in variants with functional impact may explain the ancestry-related differences observed in drug dependence GWAS. To verify this hypothesis, we analyzed two different datasets with AA and EA samples used in our recently published AD GWAS, as described above. In this study, ten GWAS-significant variants were observed associated with AD symptom counts in one ancestry group but not in the other (Supplementary Table 2 & 3) – i.e., in either AAs or EAs but not both. These genetic variants are located in ADH1B, ADH1C and PDLIM5. In PDLIM5, we observed large ΔF values for African ancestry, but the PDLIM5 significant variant (i.e., rs10031423) showed low ΔF values in all ancestry groups (Supplementary Figure 4). In ADH1B, we observed large ΔF values for Asian ancestry (i.e., ΔF>0.6), but low ΔF values of ADH1B variants for European and African ancestries (Supplementary Figure 5). In ADH1C, extreme ΔF values were observed for Asian ancestry (i.e., ΔF>0.6), and most of the Asian ΔF peaks are genome-wide significant variants for AD symptom counts (Supplementary Figure 6). Both
