In this study, our primary aim was to identify alternatively spliced exons that contribute to the susceptibility to AUD. To achieve this aim, we established a computational model to predict cis-regulated exon skipping in dorsolateral prefrontal cortex (DLPFC), using data from the CommonMind Consortium (CMC) [21]. Using the resulting prediction models, potentially causal skipped exons were identified by a Mendelian randomization (MR)-based approach that examined the association between the genotype-imputed cis-regulated components of exon skipping and DSM-IV alcohol dependence and symptom counts, based on individual genotype and phenotype data from the Collaborative Study on the Genetics of Alcoholism (COGA) [22–24]. Significant results were evaluated in a replication dataset of the Australian Twin-family Study of Alcohol Use Disorder (OZALC) [25]. In particular, an alternatively spliced non-coding exon in ELOVL7 was found to impact AUD susceptibility; its impact was further evaluated by leveraging the power of additional large-scale GWAS. Furthermore, the impact of the ELOVL7 skipped exon in the brain was visualized by association analysis with magnetic resonance imaging data from the UK Biobank. Predicted downstream genes and biological pathways of the
