In practice, we partition the genome into 1703 largely independent genomic regions estimated using data from the 1KG European sample25–27 [http://bitbucket.org/nygcresearch/ldetect-data], and conduct multivariate update of the effect sizes within each LD block (see Supplementary Note). To avoid numerical issues caused by collinearity between SNPs, we set a lower bound on the amount of regularization applied to the genetic markers (i.e., restricting \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\phi ^{ - 1}\psi _j^{ - 1} \ge \rho$$\end{document}ϕ-1ψj-1≥ρ, where ρ is a small constant). We use ρ = 1 throughout this paper.
