The discovery GWAS were from the meta-GWAS of problematic alcohol use in EA cohorts (EA-PAU) (N = 435,563) [7] and the AA GWAS of AUD from the Million Veteran Program (AA-AUD) (N = 56,648) [5], the largest GWAS of AUD-related phenotypes to date in EA and AA populations, respectively. The EA-PAU was a meta-analysis of problematic alcohol use [7] comprised of the AUD GWAS of the Million Veteran Program Phase I [5] and Phase II data, the alcohol dependence GWAS from the Psychiatric Genomics Consortium [25], and the GWAS of scores from the problem subscale of the Alcohol Use Disorder Identification Test (AUDIT items 4 to 10) in the UK Biobank [6]. Across both discovery GWAS, A/T, or C/G variants were excluded to avoid strand ambiguity. As we were focusing on AUD-associated variants shared between EA and AA, only variants having the same directions of effects in both EA-PAU and AA-AUD were included. Both GWAS summary statistics were downloaded from the database of genotypes and phenotypes (dbGaP: phs001672.v3.p1, https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs001672.v3.p1).
