Skip to main content
Fig. 4 | Journal of Cheminformatics

Fig. 4

From: The impact of cross-docked poses on performance of machine learning classifier for protein–ligand binding pose prediction

Fig. 4

The performance in terms of A top1, top2 and top3 success rates, and B binding funnel analysis when the models are trained on the PDBbind-ReDocked-Refined set and tested on CASF-Docking set. The RMSDs utilized in the original CASF benchmark are employed here. The methods colored in pink and orange are calculated in this study, while the predicted scores of the others are just copied from original paper. The averages of the random sampling of 1000 redundant copies with replacements are shown in the figure. For binding funnel analysis, the x-axis indicates the RMSD range (e.g., [0–2 Å], [0–3 Å], etc.) where the Spearman correlation coefficients between the RMSD values and the predicted scores are calculated

Back to article page