Table 4:

Performance comparisons on the SARS-CoV-2 main protease (Mpro) and epidermal growth factor receptor (EGFR) benchmark set in terms of root mean square error (RMSE) in kcal/mol, Pearson correlation coefficient (R) and Spearmann correlation coefficient (R_S) for different models before and after retraining using LP-PDBBind.

Mpro Model	RMSE		R		R S
	original	retrained	original	retrained	original	retrained
AutoDock Vina	1.20	1.17	0.55	0.66	0.51	0.68
IGN	1.86	1.44	0.64	0.61	0.69	0.65
RF-Score	2.06	1.64	0.43	0.52	0.47	0.58
DeepDTA	1.18	1.12	0.59	0.63	0.60	0.66
EFGR Model	RMSE			R		R S
	original	retrained	original	retrained	original	retrained
AutoDock Vina	3.11	1.59	0.25	0.38	0.21	0.36
IGN	1.06	0.96	0.36	0.65	0.17	0.62
RF-Score	1.57	0.97	−0.15	0.52	−0.18	0.45
DeepDTA	1.22	1.38	0.23	0.06	0.20	0.23