Fig. 3: Comparison of the performance of MCCP and the empirical method on complex disease risk prediction using PRS as predictor. | Nature Communications

Fig. 3: Comparison of the performance of MCCP and the empirical method on complex disease risk prediction using PRS as predictor.

From: Translating polygenic risk scores for clinical use by estimating the confidence bounds of risk prediction

Fig. 3

For MCCP, sample coverage (x axis) indicates the proportion of samples predicted as cases or controls, whereas, for the empirical method, it indicates extreme PRS, e.g., top and bottom x% of PRS. AUCs are computed from multivariate logistic regressions adjusted for age, sex, and PC1–6 on these samples stratified from MCCP and empirical method, respectively. The expected error rates for MCCP are indicated by the size of data points up to 0.20. Vertical lines correspond to an expected error of 0.05 from the MCCP. The solid lines and shades represent the median and 95% confidence intervals of AUCs. CAD coronary artery disease, T2D type 2 diabetes mellitus, IBD inflammatory bowel disease, BRCA breast cancer, SCZ schizophrenia, T2D (MDC) T2D data set from the MDC study at the baseline. Source data are provided as a Source Data file.

Back to article page