current analyses). Thus, the derivation of PRS did not involve a completely independent discovery and target dataset. As an overlap between these datasets may potentially lead to an overestimation of the prediction accuracy for the PRS (Wray et al., 2013), interpretations regarding the main effect of the PRS on heavy episodic drinking outcomes should be made with some caution. On the other hand, it is possible that this estimation may simply reflect the enrichment of genes for alcohol use in both samples. In other words, given that alcohol use disorders are known be influenced by many genes of small effect, using genetically-related case-control samples that are likely enriched for the same small-effect genes associated with alcohol use may lead to more robust characterizations of risk. Although this is beyond the scope of the current study, future studies should examine whether PRS effects significantly differ between discovery samples with related versus and non-related individuals to the target sample. Third, heavy episodic drinking was defined as having five or more alcoholic drinks within a 24-hour period, which is a more stringent threshold for females where four drinks are considered the typical threshold. Hence, sex differences regarding genetic and environmental influences on trajectories
