Polygenic risk scores in human genetics for study design discovery and translation

Abstract

Polygenic risk scores (PRS) quantify the component of disease risk captured by measured additive common variants. Their use as a study design variable, rather than only as a predictive endpoint, broadens their scientific value in human genetics. PRS can define informative extremes of common-variant burden, identify discordance between phenotype and PRS-estimated risk, and facilitate the detection of subgroup-specific or residual mechanisms that may be obscured in conventional case-control analyses. This review outlines four major PRS-informed design strategies: tail sampling based on PRS extremes; discordance sampling based on mismatch between phenotype and PRS-implied risk; conditional and stratified genome-wide association analyses using PRS to adjust for / partition background risk; and residual phenotype analysis of the component of phenotype remaining after the PRS-associated component has been removed. It thus provides a practical framework for sample enrichment, subgroup definition, and calibrated epidemiologic comparison. These designs may be especially informative when integrated with sequencing, multi-omics, longitudinal cohorts, and translationally oriented intervention studies. Their application requires careful attention to data leakage, ancestry-related bias, collider structures, and the limits of interpretation.