We quantified the accuracy of multiple polygenic scores (PGSs) for height on the basis of GWS SNPs (hereafter referred to as PGSGWS) and on the basis of all HM3 SNPs (hereafter referred to as PGSHM3). PGSGWS were calculated using joint SNP effects from COJO, and PGSHM3 using joint effects calculated using the SBayesC method28 (Methods). We denote \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R}_{{\rm{GWS}}}^{2}$$\end{document}RGWS2 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R}_{{\rm{HM}}3}^{2}$$\end{document}RHM32 as the prediction accuracy of PGSGWS and PGSHM3, respectively. For conciseness, we also use the abbreviations PGSGWS-X and PGSHM3-X (and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R}_{{\rm{GWS}}-{\rm{X}}}^{2}$$\end{document}RGWS−X2 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R}_{{\rm{HM}}3-{\rm{X}}}^{2}$$\end{document}RHM3−X2) to specify which GWAS meta-analysis each PGS (and corresponding prediction accuracy) was trained from. For example, PGSGWS-METAFE refers to PGSs based on 12,111 GWS SNPs identified from our METAFE.
