To date, GWAS have been overwhelmingly limited to individuals of primarily European descent (Mills & Rahal, 2019). Because of variation in allele frequencies and linkage disequilibrium (LD) patterns, PGS often lose predictive accuracy when there is mismatch between the genetic similarity of the discovery GWAS and target sample (Ding et al., 2023). COGA includes participants of both African-like and European-like groupings, thus we used PRS-CSx (Ruan et al., 2022), a method that integrates GWAS summary statistics from well-powered GWAS (typically of European-like individuals) with those from other populations to improve the predictive power of PGS in the participants of African-like groupings in COGA. PRS-CSx employs a Bayesian approach to correct GWAS summary statistics for the non-independence of SNPs in LD. We converted PGS into Z-scores for ease of interpretation
