In order to obtain weights for constructing polygenic scores, summary statistics files (i.e., GWAS results) were clumped to include only variants in approximate linkage equilibrium, using second generation PLINK49,50 and the following thresholds (varying the linkage disequilibrium r2 threshold): --clump-kb 500, --clump-p1 1, --clump-p2 1, --clump-r2 0.2 (and .05 and .01). 1000Genomes data were used as the source for linkage disequilibrium information for pruning all summary statistic files, and we varied the source populations to include all samples, and each of the major populations individually.
