characteristic of AUD or could have resulted from it; they may not reflect etiology of AUD. However, our approach could have broad importance as a follow-up to well-powered genetic studies by identifying variants that have functional effects on gene expression within the usually broad loci identified in such studies. Finally, ASE analysis from RNA-seq data is subject to technical challenges due to the differences in our ability to align sequence reads to two alleles. Most alignment algorithms more readily align the sequencing reads with the reference sequence, which can lead to depressed expression signals for the alternative allele. Fortunately, this potential bias can be avoided by comparing the allelic imbalances between two experimental conditions, since the biases due to alignment algorithms will be the same in both. Thus, we focused our analysis on the differences in allelic imbalance between subjects with and without AUD. This strategy also allowed us to focus our analysis on the genetic variants that were associated with AUD.
