This study has several limitations. First, COGA is a family cohort enriched with AUD cases; thus, our findings may not be widely generalizable. Second, our FH− subsample size was small, therefore, limited statistical power likely contributed to the null findings in that group. Third, we limited our analysis to European ancestry samples only, due to the small sample sizes and therefore limited power in both the discovery and target datasets, as well as complicated LD patterns in admixed populations such as African American and Hispanic populations. Fourth, we adjusted for sex and birth cohorts in our analysis but suspect that there are other important family and environmental factors not evaluated. Fifth, PRS-CS needs an external LD reference panel. Although it is relatively insensitive to the different LD structures between the external reference panel and experimental datasets used, these differences could still potentially cause problems (Ge et al., 2019). Lastly, the PRS we used was from a meta-analysis of MVP-AUD (Kranzler et al., 2019) and UKBB-AUDIT-P (Sanchez-Roige et al., 2019). As noted, these are from somewhat different although correlated phenotypes (Sanchez-Roige
