Polygenic scores for MDD and loneliness were computed using the PRSice software package [32] (Supplementary Methods) for each individual in BioVU and ARIC as follows:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{Polygenic}}\;{\mathrm{score}} = \mathop {\sum }\limits_{i = 1}^{\# SNPs} (wi\;{\mathrm{x}}\left[ {{\mathrm{SNP}}\;{\mathrm{genotype}}} \right]i)$$\end{document}Polygenicscore= ∑i=1#SNPs(wixSNPgenotypei)where the SNP genotype was coded as 0, 1 or 2 and wi was the ß coefficient (representing effect size of the marginal association between the SNP and the phenotype) from the Psychiatric Genomics Consortium meta-GWAS of MDD [11], or the Lonely Consortium meta-GWAS of loneliness [10]. We selected SNPs into the polygenic score if their association P in the meta-GWAS was below a specified threshold. In the phenome-wide association study, we selected an P threshold of 0.05, a value previously found to generate scores that maximized the prediction of MDD [11] and other psychiatric traits [33], and included 392,372 and 379,906 SNPs in the polygenic scores of MDD and loneliness, respectively. In targeted analyses of CAD, we selected the P threshold at which the polygenic score was most strongly associated with CAD risk (i.e., at
