We ran Genome-wide Complex Trait Analysis (GCTA, [42]) to estimate the amount of variance tagged by SNPs on the Illumina 550K platform in the total sample. In addition, we assessed the predictive utility of polygenic risk scores by splitting the sample in half, and using one half as the discovery sample and the other half as the replication sample. These analyses utilized the polygenic risk score method available in Plink [43, 44], employing p-value cutoffs of 0.1- 0.5 in increments of 0.1.
