Figure 1

C-score threshold. The graph on the left shows the discrimination ability of C-scores for variants annotated as either benign or pathogenic by ClinVar (random sample of 6,000 of each type of variant from the ClinVar database edition 20150305). Considering only variants with C-score above a threshold: proportion that retained benign variants in blue versus pathogenic in red. The dashed crossing lines indicate our threshold, 4.075, chosen to capture 50% of pathogenic variants and minimal benign variants (3.2%). The graph on the right shows the percentage of benign variants included versus the percentage of pathogenic variants included as the C-score threshold varies. The curve approaching the upper left indicates that combined annotation-dependent depletion is a good discriminator of pathogenic versus benign (area under curve = 89.7%).