One limitation of the current study is that, since it was designed to answer the general question whether RNA splicing impacts the genetics of AUD, it cannot answer cell type-specific questions. Our study design used bulk RNA-seq data because the statistical genetics analysis required the power of large-scale cohorts of samples with known genotypes, AUD-related phenotypes, and transcript-level quantifications. In addition, splicing analysis requires sufficient read depth that is not currently available in single-cell RNA sequencing data. Moreover, the relatively small sample size such as COGA and OZ-ALC may limit the power in discovering significant splicing events. Another limitation is that causality cannot be directly verified because of the challenges in experimentally modeling complex traits such as AUD, as cell culture studies and animal models cannot completely represent the human disease system. Nevertheless, a major strength of our study is that the use of the MR methodology, including the GSMR, leverages high power in causality inference from large-scale datasets. This method enables identification of causal splicing events, which provides new information on the role of RNA splicing in AUD risk.
