The PRS was constructed with a clumping-and-thresholding approach with the first five principal components added as covariates (see Methods). The PRS constructed with the best performing P value threshold (PT = 0.2, which accounts for 24,034 independent variants) explained ~5% of PTSD onset variance (Nagelkerke R2 = 4.68%, P = 0.003) as well as PTSD severity variance (R2 = 4.35%, P = 0.0008) as measured by CAPS (a quantitative measure of PTSD severity that ranges from 0 to 120).
