Table 3:

Performance comparisons on the BDB2020+ benchmark set in terms of root mean square error (RMSE) and Pearson correlation coefficient (R) for different models before and after retraining using LP-PDBBind.

Model	RMSE (kcal/mol)			R
	original	retrained	difference	original	retrained	difference
AutoDock Vina	3.29	2.23	−32%	0.21	0.28	+33%
IGN	1.61	1.38	−14%	0.38	0.52	+37%
RF-Score	1.84	1.59	−14%	0.31	0.50	+61%
DeepDTA	1.99	2.15	+8%	0.17	0.06	−65%