PRS is often trained on GWAS summary statistics data, without the need to directly access raw individual-level genotype training data, which is often not readily available. Reliable estimation of PRS parameters, however, requires a large sample size GWAS. GWAS summary statistics data, unlike individual-level genetic data, are often publicly available for many traits and diseases. It typically contains results from univariate association test statistics on a variant-per-row format. In this study, GWAS summary statistics data from two studies were used as base/discovery data sets:1 PTSD GWAS summary statistics data, which is the largest published PTSD study to date (n = 10 k, 25% cases) with European ancestry participants17, and2 Schizophrenia GWAS summary statistics data, which is the largest psychiatric genetic study (37 K cases and 113 K controls) to date, also consisting of mostly European ancestry participants27.
