BgN-Score and BsN-Score: Bagging and boosting based ensemble neural networks scoring functions for accurate binding affinity prediction of protein-ligand complexes

Table 2 Comparison of the scoring powers of BsN-Score, BgN-Score, SNN-Score, Random Forests (RF), and 16 conventional SFs on the core test set Cr

Scoring function	N ¹	R _p ²	R _s ³	SD ⁴	RMSE_test⁵	RMSE_train⁶
BsN-Score::XARG	195	0.816	0.799	1.38	1.386	1.366
BgN-Score::XARG	195	0.804	0.798	1.42	1.449	1.403
RF::XARG	195	0.801	0.790	1.43	1.498	1.442
SNN-Score::X	195	0.675	0.685	1.76	1.760	1.704
X-Score::HMScore	195	0.644	0.705	1.83	1.865	1.730
DrugScore^CSD	195	0.569	0.627	1.96	-	-
SYBYL::ChemScore	195	0.555	0.585	1.98	-	-
DS::PLP1	195	0.545	0.588	2.00	-	-
GOLD::ASP	195	0.534	0.577	2.02	-	-
SYBYL::G-Score	195	0.492	0.536	2.08	-	-
DS::LUDI3	195	0.487	0.478	2.09	-	-
DS::LigScore2	193	0.464	0.507	2.12	-	-
GlidScore-XP	178	0.457	0.435	2.14	-	-
DS::PMF	193	0.445	0.448	2.14	-	-
GOLD::ChemScore	178	0.441	0.452	2.15	-	-
SYBYL::D-Score	195	0.392	0.447	2.19	-	-
DS::Jain	189	0.316	0.346	2.24	-	-
GOLD::GoldScore	169	0.295	0.322	2.29	-	-
SYBYL::PMF-Score	190	0.268	0.273	2.29	-	-
SYBYL::F-Score	185	0.216	0.243	2.35	-	-

¹ Number of complexes in Cr with positive (favorable) binding scores using this SF [14].
² R_p is the Pearson correlation coefficient between predicted and measured BA values of complexes in Cr.
³ R_s is the Spearman correlation coefficient between predicted and measured BA values of complexes in Cr.
⁴ SD is the standard deviation of errors between predicted and measured BA values of complexes in Cr based on Equation 3 in [15].
⁵ RMSE is the root-mean-square of errors between predicted and measured BA values of the test complexes in Cr. Test RMSE is not available for conventional SFs except for X-Score::HMScore that we have re-constructed.
⁶ RMSE is the root-mean-square of errors between predicted and measured BA values of out-of-sample complexes in the training set Pr. Training RMSE is not available for conventional SFs except for X-Score::HMScore that we have re-constructed.

ISSN: 1471-2105