R2 of the PRS was computed as the difference between the R2 of full model that contains PRS along with covariates and the R2 of null model that contains only covariates. Similarly, Nagelkerke’s pseudo-R2 of the PRS was computed as the difference in Nagelkerke’s pseudo-R2 between that of the full model and null model. The p value of the null hypothesis that the regression coefficient of PRS is zero is reported with the R2. Odds ratios and difference in mean CAPS (and corresponding 95% confidence intervals) between quantiles were computed using the glm function in R (with family = “binomial” or family = “gaussian”), with the first five principal components added as covariates to control for population stratification. The power analysis for single variants in the Introduction section (for the previous largest GWAS) was done using GAS power calculator. Standardization was done by converting raw scores to z scores (centering by mean and scaling by standard deviation). All computations were done on R (version 3.2.3) statistical computing environment and PLINK (v1.90b3s 64-bit)23.
