Genetic determinants of polygenic prediction accuracy within a population

<strong>Full summary statistics</strong> of genome-wide discovery of prediction variability quantitative loci using Levene's test, for 15 quantitative biomarkers <br> <strong>Scripts</strong> <strong>pipeline.sh:</strong> command line codes used for analyses <strong>dominance_interaction.R:</strong> R script for identifying dominance and interaction effects <strong>dominance_PRS.R:</strong> R script for incorporating dominance effects into polygenic risk scores <br> <strong>Supplementary Tables</strong> <strong>Table S1:</strong> Cohort characteristics of unrelated UK Biobank European ancestry participants. <strong>Table S2:</strong> Summary of environmental and lifestyle exposures in the UK Biobank discovery set (N = 254,376). <strong>Table S3:</strong> Proportion of phenotypic variance explained by polygenic risk scores in the UK Biobank. Proportion of variance explained (adjusted R2) was assessed by linear regression. <strong>Table S4:</strong> Summary of all COJO-independent SNPs used for constructing polygenic risk scores. Results of COJO analyses are included. <strong>Table S5:</strong> Summary of all COJO-independent PVL SNPs. Results of COJO analyses are included. <strong>Table S6:</strong> Annotation of PVLs. <strong>Table S7:</strong> Summary statistics of genetic correlation estimates based on GWAS-identified risk loci and PVLs respectively. <strong>Table S8:</strong> Summary of dominance effects and SNP-exposure interaction effects tested for COJO-independent SNPs, and GWAS risk loci-PVL colocalization. <strong>Table S9:</strong> Summary of dominance effects and SNP-exposure interaction effects tested for COJO-independent PVL SNPs. <strong>Table S10:</strong> Additional dominance effects with an FDR &lt; 0.05 combined with polygenic risk scores. Coefficients obtained from multivariate linear regression are provided. Polygenic risk scores were standardized to have zero mean and unit variance. <strong>Table S11:</strong> Additional dominance effects with Bonferroni-corrected p-value &lt; 0.05 combined with polygenic risk scores. Coefficients obtained from multivariate linear regression are provided. Polygenic risk scores were standardized to have zero mean and unit variance.