(Table S1), and 1 061 130 of 1 126 428 variants were present in Indiana Biobank after QC; both PRSgene and PRSall had P-values < 0.05 (PRSgene Beta=0.11, SE = 0.02; PRSall Beta = 0.34, SE = 0.05) but not PRSintergenic (Beta = 0.02, SE = 0.05). Results of using different window sizes to extend gene boundaries are in Table S4 and Fig. S1. The numbers of variants increased slightly with larger window sizes, and windows 50 and 100 Mb had the same number of variants, indicating that most variants are located within or close to genes. Overall, the results were similar, therefore, we kept variants within gene boundaries because it was more straightforward to determine AUD genes, as larger distances often contained multiple genes, and it is challenging to assign intergenic variants to a gene.Table 2Associations between AUD and PRSgene, PRSintergenic, and PRSall in AA and EA.PRSgenePRSintergenicPRSallPopulationTarget datasetBetaSEP-value# VariantsBetaSEP-value# VariantsBetaSEP-value# VariantsAAAlla0.170.033.55E−088580.030.030.326750.120.039.42E−051,126,428COGA0.150.049.67E−048580.020.040.616750.060.040.161,126,428SAGE0.180.076.27E−038580.120.070.106750.180.070.011,126,428YalePenn0.210.057.61E−05858−0.020.060.766750.170.067.28E−031,126,428EAIndiana Biobank0.110.050.028470.020.050.596660.340.052.35E−211,061,130Significant P-values are in bold.PRSgene PRS calculated using concordant variants located in genes associated with AUD in both AA and EA, PRSintergenic PRS calculated using concordant variants located outside genes associated with AUD in both AA and EA, PRSall PRS calculated using all variants.aCOGA, SAGE, and YalePenn combined.
