We tested whether genetic risk for alcohol consumption is predictive of differences in postmortem gene expression. Precomputed gene-expression RNA-sequencing weights for 9 brain regions and the liver from the GTEx project (30) were analyzed using the FUSION suite (27). Analyses used GWAS results for alcohol consumption from the UK Biobank (28). Results were Bonferroni-corrected for n = 9839 tests across the 10 tissues (Supplemental Data). Replication was tested using an independent alcohol-consumption GWAS (N = 70,460) (29) and dorsolateral prefrontal cortex (DLPFC) gene-expression weights from the CommonMind Consortium (31). As the gene that showed the strongest association in the discovery dataset was not present in the replication data, we examined whether any of the Brodmann area 9 (BA 9) gene-expression associations at FDR-corrected p < .05 were significant in the replication data (see Results). Replicated genes were probed for association with other GWAS phenotypes using a phenome-wide association study (PheWAS) implemented through the “GWAS Atlas” browser (65). BA 9 in the GTEx dataset and DLPFC in the CommonMind consortium dataset overlap with the prefrontal regions implicated in our neuroimaging analyses (i.e., middle and superior frontal gyri) (see Results). No postmortem insula data were available.
