PRS-CARV: A summary statistics framework for integrating annotation-informed rare variants to improve polygenic risk...

Polygenic risk scores (PRS) primarily rely on common variants because rare variant effects are difficult to estimate without large sequencing cohorts. We developed PRS-CARV, a summary statistics framework that integrates annotation-informed rare variant burden scores with common-variant PRS. In simulations, PRS-CARV consistently outperformed common-variant PRS alone. Applied to whole-genome sequencing data from the nuMoM2b-Heart Health Study, incorporating rare variants improved prediction of lipid traits by 2-11% compared with common-variant PRS. PRS-CARV enables practical integration of publicly available rare variant summary statistics, improving polygenic prediction and supporting broader implementation of precision medicine.