Factors influencing the power of polygenic risk score: a survey based on a genotype array dataset of an ulcerative colitis cohort

Abstract

BackgroundPolygenic risk scores (PRSs) are personalized assessments based on the cumulative effect of a larger number of genetic variants, and the PRS has been utilized to estimate the risk of a disease or disease-related trait for an individual. However, the capability of PRS can be interfered with by multiple factors spanning from the generating genome-wide association study (GWAS) summary to the PRS calculation. In this study, based on a large cohort of 2,781 ulcerative colitis (UC) cases and 2,876 healthy controls with available genotype array data, we systematically analyzed factors that could affect the power of PRS, including GWAS summary statistics, population stratification, impact of variants and imputation reference panels.ResultsWe eventually established the best model achieving an AUC of 0.715 (95% CI 0.702-0.729) in prediction using 2,669 UC cases and 2,846 healthy controls of genetically identified Europeans. And samples in the top 5% of the PRS distribution had more than seven times higher risk of UC than samples in the lowest 95% (OR=7.014, 95% CI 4.977-9.884).ConclusionOur analyses demonstrated that including population-specific, more associated SNPs and selecting GWAS summary statistics results from similar population structure can improve the power of PRS, strictly following the principle of focusing on one population in all aspects of generating PRS, which can be a cost-effective way to apply genotype array-derived PRS to practical risk estimation.