LAUSR

Background Polygenic scoring has emerged as one way to characterize aggregate genetic risk; however, the conventional methods for calculating polygenic scores contain a mixture of “true” genetic signal and random noise. We hypothesized that functional genomic information could be used to enhance polygenic signal to predict young adult alcohol use, and to identify genetic variants (single nucleotide polymorphisms; SNPs) likely to be enriched for gene-by-environment interaction. In polygenic analyses in the FinnTwin12 sample, variants located under a DNase I peak or in linkage disequilibrium with a SNP under a DNase I peak (i.e., in an open chromatin region and likely to have a regulatory function) had per-SNP effects that were > 3.5 times higher than non-DNase SNPs. Furthermore, DNase SNPs were enriched for gene-by-environment interaction compared to SNPs filtered by p-value only (p = .047) in tests where romantic relationship status was the environmental moderator. This project examines the use of functional annotation information to improve polygenic risk prediction and the implications for studies of measured gene-environment interaction.