Sfcnn: a novel scoring function based on 3D convolutional neural network for accurate and stable protein–ligand affinity prediction

Table 3 Comparison between Sfcnn and the top 10 scoring functions tested on the CASF-2016 benchmark

Scoring function	R	SD	Size	Description
ΔVinaRF₂₀	0.816	1.26	285	Machine learning
Sfcnn	0.792	1.32	283	Machine learning
X-Score	0.631	1.69	285	Empirical
X-ScoreHS	0.629	1.69	285	Empirical
ΔSAS	0.625	1.7	285	Single descriptor
X-ScoreHP	0.621	1.7	285	Empirical
ASP@GOLD	0.617	1.71	282	Knowledge-based
ChemPLP@GOLD	0.614	1.72	281	Empirical
X-ScoreHM	0.609	1.73	285	Empirical
AutoDockVina	0.604	1.73	285	Empirical
DrugScore2018	0.602	1.74	285	Knowledge-based

Results (excluding Sfcnn) cited from Su et al. [14]. The performance of these scoring functions was recalculated by us for comparison

ISSN: 1471-2105