Figure 3

Superiority over existing methods.
Rank receiver operating characteristic curves (ROCs) of snvForest, eXtasy and SPRING in validation experiments against (A) neutral, (B) HGMD and (C) combined test samples. Curves are drawn by pooling the validation results for all the 2,511 diseases and those “plots within plots” are zoomed-in regions of AUC curves. The curves of snvForest and SPRING both stay above that of eXtasy, indicating the higher ability of the former two methods to enrich true causative variants among top rank positions.