Skip to main content

Table 3 Measured accuracies of eight in silico predictors as benchmarked against seven different variant reference datasets. Measured accuracies are calculated as the areas under the respective ROC curves (AUCs) and Matthews correlation coefficients (MCCs). See Additional file 1: Figure S4 for the ROC curve graphs

From: Variant effect prediction tools assessed using independent, functional assay-based datasets: implications for discovery and diagnostics

  ClinvarHC Humsavar Swissvar Varibench TP53-TA BRCA1-DMS UniFun
  AUC MCC AUC MCC AUC MCC AUC MCC AUC MCC AUC MCC AUC MCC
GERP++ 0.863 0.587 0.777 0.469 0.677 0.286 0.571 0.15 0.719 0.283 0.544 0.069 0.538 0.04
fitCons 0.641 0.3 0.533 0.033 0.564 0.008 0.651 0.024 0.557 0 0.559 0 0.515 0.033
SIFT 0.848 0.489 0.841 0.543 0.698 0.289 0.651 0.228 0.835 0.484 0.653 0.199 0.631 0.184
PolyPhen 0.827 0.447 0.831 0.541 0.699 0.301 0.672 0.256 0.859 0.469 0.596 0.088 0.623 0.168
CADD 0.939 0.731 0.851 0.57 0.73 0.331 0.663 0.25 0.869 0.418 0.556 0.032 0.589 0.119
Condel 0.879 0.51 0.911 0.664 0.728 0.333 0.86 0.57 0.883 0.074 0.747 0.172 0.614 0.098
REVEL 0.945 0.68 0.968 0.83 0.792 0.462 0.89 0.59 0.907 0.465 0.737 0.088 0.63 0.148
fathmm 0.787 0.288 0.902 0.538 0.701 0.253 0.936 0.509 0.53 0 0.621 0 0.531 0.02