of ten: PDLIM5 rs10031423, ADH1B rs1693457 and ADH1C rs6846835. Specifically, PDLIM5 rs10031423 showed R2 = 0 with respect to ADH1B rs1693457 and ADH1C rs6846835 in AAs and EAs, whereas ADH1B rs1693457 and ADH1C rs6846835 showed R2 = 0 in EAs and R2 = 0.15 in AAs. Slight differences exist for these variants in the present association results compared with the published AD GWAS (Supplementary Table 3), because in the present study we used the original SAGE-imputed data, and symptom counts instead of substance dependence diagnosis to adjust for multiple dependencies. To estimate the effect of common variants on the genome-wide significant associations with Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV) symptom counts for AD, among the common variants investigated in the first part of the study, we selected those with ΔF>0.10 in African or European ancestries that are present in Yale–Penn and SAGE datasets (12,969 variants for African ancestry, and 8721 variants for European ancestry). We chose this threshold in order to exclude those variants with minimal allele ΔF among human populations. Then, performing separate analyses for AAs and EAs and for the Yale–Penn and SAGE datasets, we estimated the association of rs10031423, rs1693457 and rs6846835 with
