be at least partially due to small sample sizes of COGA AA and the AA discovery dataset. Another limitation is that participants in Indiana Biobank were not ascertained to study AUD and related traits. All data were obtained from EHR, removing the ability to determine remission types, AUD severity, treatment history, and family history of remission, all of which limits our ability to perform detailed analyses. Lastly, to test our hypothesis, we used PGS derived using AUD GWAS as the discovery datasets, which may be different from the remission datasets used in this study.
