To improve the performance of PRS, more disease-associated variants should be included and unrelated variants should be excluded. AUD is caused by many genes with small effects and in GWAS of AUD, due to the large number of variants tested, many variants that are unrelated to AUD show some degree of association (e.g., P-values < 0.05) purely by chance (i.e., false positives). If sample sizes are large (e.g., hundreds of thousands of participants or more), while the majority of AUD-associated variants are still not genome-wide significant, they usually have smaller P-values than those false positives and can still contribute to the calculation of PRS. However, when the discovery GWAS sample sizes are small to moderate, the discrimination between AUD-related and unrelated variants narrow. This leads to a reduction in PRS performance. Using large-scale EA discovery GWAS could mitigate this problem, but the improvement is limited even with sophisticated statistical methods due to the differences between the discovery GWAS and the target datasets [23]. Our gene-based PRS framework leverages the concordant variants across different populations and discriminates variants unrelated to the
