The estimated PRSs were constructed from GWASs of the simulated genotypes (modeled after chromosome 20) in European and African ancestry populations, each with 10,000 cases and 10,000 controls. Odds ratios (ORs) were estimated for all variants with minor allele frequency (MAF) > 1%, and statistical significance of association was assessed with a chi-square test. While causal variants may be included in the estimated PRS, they are drawn from the total allele frequency spectrum; thus, they are primarily rare (93% and 87% of causal variants have MAF < 1% in European and African ancestry populations when m = 1,000) and restricted from our analysis. For each population, variants were selected for inclusion into the estimated PRS by p value thresholding (p = 0.01 [main text], 1 × 10−4, and 1 × 10−6 [Supplemental information]) and clumping (r2 < 0.2) in a 1 Mb window within the GWAS population, where r2 is the squared Pearson correlation between pairs of variants. A fixed-effects meta-analysis was also performed to calculate the inverse-variance weighted average of the ORs in African and European ancestry populations, and LD r2 values for clumping used both datasets as the reference.
