Efficient estimation and applications of cross-validated genetic predictions

This study presents a novel reference-free method to produce polygenic risk scores (PRS) that does not rely on an external cohort, such as required matching on ancestry and genotyping platform or imputation quality.

This reported algorithm avoids both over-fitting and prohibitive increases in computational time. Here, researchers demnostrate several novel applications of this new reference-free PRS including detection of pleiotropy across 246 metabolic traits.