Four different genotyping arrays were used: the Illumina 1M, Illumina OmniExpress 12VI, and Illumina 2.5M (Illumina, San Diego, CA), and Smokescreen (BioRealm LLC, Walnut, CA). Quality control and imputation procedures are described in Lai et al.29. Imputed genotypes for the COGA parent-offspring trios, in combination with summary statistics from independent GWAS discovery samples, were used to construct the genome-wide polygenic risk scores using PRS-CSx41. This procedure uses ancestry-specific discovery sample GWAS weights, paired with linkage disequilibrium information from an ancestry-matched external reference panel, to estimate the posterior effect size for each SNP. Reference panels from the 1000 Genomes Phase III European or African subsamples were used for the EA and AA groups, respectively. For participants of European ancestry, the discovery sample consisted of a meta-analysis of DSM-IV alcohol dependence from the Psychiatric Genomics Consortium (COGA sample removed)42, AUDIT-P from the UKBiobank43, and DSM-5 Alcohol Use Disorder from the Million Veteran Program in individuals of European ancestry44. For participants of African ancestry, GWAS summary statistics were drawn from the meta-analyzed European ancestry summary statistics in tandem with the African ancestry GWAS
