Figure 4

Comparison of SuRFR, GWAVA, CADD and FunSeq on two ClinVar datasets. (A,B) ROC curves (true positive rate versus false positive rate) and AUCs for SuRFR, GWAVA, CADD and FunSeq run on ClinVar pathogenic versus non-pathogenic variants (A) and ClinVar pathogenic versus matched 1000 Genomes background variants (B). SuRFR outperforms all three methods on both of these datasets, with AUCs of 0.802 and 0.846 versus 0.705 and 0.802 for GWAVA, 0.763 and 0.831 for CADD and 0.544 and 0.483 for FunSeq on the two datasets, respectively.