UCLA

Polygenic risk score (PRS) is a method that utilizes the effect sizes of genetic variants on a particular disease or trait to evaluate an overall genetic risk for a certain individual. Such effect sizes are often estimated using traditional genome-wide association study (GWAS) for the trait of interest. There are methods developed that aim to improve the predictive power of PRS by incorporating the genetic information from multiple related traits. One existing popular method is MTAG, which requires GWAS summary statistics from multiple traits and is based on strong assumptions about genetic correlation across traits. We developed some variations of MTAG and evaluated their performance for computing PRS against GWAS, using a variety of trait data from UK Biobank as well as simulated data.