Figure 4

Recovery of DUD actives using various fingerprints. (A) Average AUC values and (C) enrichment factors at 5% (EF5%) for recovery of 40 sets of actives in directory useful decoys (DUD) from the corresponding decoys set by various fingerprints, using CBD _fingerprint (violet bars) and T _fingerprint (grey bars) as scoring functions. (B) AUC values and (D) EF0.1% values for recovery of DUD actives from the entire ZINC database. (E) Occupancy heat map of the molecular shape triangle by DUD actives and decoys (128,352 cpds, blue ≤ 2 cpd/pixel to magenta ≥ 150 cpds/pixel) and (F) by the entire ZINC database (23.2 M cpds, blue ≤ 50 cpd/pixel to magenta ≥ 10000 cpds/pixel). See Additional file 1: Table S1-S8 for detailed AUC and EF values and Additional file 1: Figure S4-S7 for ROC curves.