For all samples, PRS were coded for every individual by multiplying an individual’s number of effect alleles at a particular SNP by that SNP’s effect size (beta) from the discovery GWAS (Sanchez-Roige et al., 2018) then averaging across SNPs to create one score per person. The discovery GWAS summary statistics were clumped using PLINK (Chang et al., 2015; Purcell et al., 2007) using the linkage disequilibrium (LD) pattern from the 1000 Genomes European (Consortium, 2015) reference sample, with an LD threshold of r2≥0.25 and a 500 kb physical distance. PRS were constructed for 8 thresholds (pT<0.0001, pT<0.001, pT<0.01, pT<0.10, pT<0.20, pT<0.30, pT<0.40, pT<0.50) and standardized using the scale function in R within each sample to ease interpretation of effect sizes. For ALSPAC and COGA, PRS were generated using PLINK; for UKB and GS samples, PRS were created using PLINK implemented in PRS-ice2 (Euesden et al., 2014) using the same R2 thresholds for clumping and the same p-value thresholds.
