A GWAS on DSM-IV alcohol dependence symptoms was first conducted in the adult European-American subsample (age 29–88; n=1,249; 45% male) of the original COGA study (see Wang et al., 2013 for a similar analysis using this sample). This sample was genotyped on the Illumina Human OmniExpress array 12.VI. 587,378 SNPs with minor allele frequency >5% were analyzed. Covariates included sex, age at interview, and cohort. Manhattan and Q-Q plots from the GWAS are provided in the Appendix section (Supplemental Figures 1 and 2, respectively). From this GWAS, “weights” were derived from t-scores of the most significant SNPs based on various p-value thresholds, from p-values .05 to .50, which were then used to calculate a PRS for participants in the Prospective Study using PLINK software (Purcell et al., 2007). For each SNP, the GWAS-derived t-scores of the associated allele were multiplied by 0, 1 or 2 (depending on the number of risk alleles that individual carried). The final PRS was the sum of the weighted SNPs. Figure 1 shows the proportion of variance explained (in sample size adjusted r2) in the
