Fig. 4

Predicted enrichment vs docking enrichment on the external validation set. The figure presents the comparison of docking enrichment in blue and predicted enrichment in red whereas the grey line in the figure represents random predictions. The comparison was used to evaluate the performance of CPVS models. The docking enrichment was created by comparing docking scores of well known inhibitors and docking scores of an external validation. Similarly the predicted enrichment was created by comparing predicted p-values for well-known inhibitors and the external validation set. AUC was also calculated and reported in the figure for comparison. Overall the CPVS models performed well and predicted enrichment is comparable to docking enrichment, except for receptor with PDB-ID 1B8O, when the predicted enrichment is a little worse than docked enrichment. The reason could be less number of known inhibitors in the top scored molecules, seen in the left bottom corner of the 1B8O graph