We calculated the deterministic accuracy in the target populations based on selected GWS SNPs. Since we focused on in this study the contribution of LD and allele frequency differences between populations to the RA, the term \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rho _b^2h_2^2/h_1^2$$\end{document}ρb2h22/h12 equalled to 1 in the simulations. We then used a reference panel to calculate the LD correlation and MAF between populations. To first validate our theory, we used Eq. (1), which assumes causal variants to be known, to calculate the LD correlation and MAF using SNP pairs between PGS-SNPs and known causal variants. We then explored the performance of our heuristic method using Eq. (2), given that causal variants are typically unknown or unobserved. For that, we took advantage of the LD and MAF information between PGS-SNPs and candidate causal variants instead. Further, we applied a naive approach assuming PGS-SNPs to be the causal variants, thus mainly the allele frequency differences between populations would be captured using Eq. (1).
