kb downstream of the transcription end site; annotated using ANNOVAR [45] based on NCBI RefSeq GRCh37) were used. To test whether using any concordant variants regardless of location would do as well, we calculated PRS using concordant variants located outside gene boundaries (referred to as PRSintergenic). We also tested whether further extending gene boundaries used to calculate PRSgene improved results by setting different window sizes: 10 kb, 25 kb, 50 kb, 100 kb, 250 kb, 500 kb, 1 Mb, 50 Mb, and 100 Mb. Lastly, we also used all variants across the entire genome to calculate PRS (PRSall) for comparison purposes. For all AA target datasets, PRSgene, PRSintergenic, and PRSall were calculated using exactly the same sets of variants, respectively, thus they were directly comparable and can be combined. PLINK [37, 38] was used to calculate PRS using the posterior effect sizes estimated by PRS-CSx and imputation dosages. All PRS were standardized as mean = 0 and standard deviation = 1 in AA (all three datasets combined) and EA target datasets separately.
