Table 1.

Machine learning scoring functions.

ML Algorithm	Name	Input Features	Dataset	Year
RF	RF-score [99]	Protein–ligand atom-type pair counts in predefined distance cutoff	PDBbind v2007	2010
	SFCscoreRF [100]	Descriptors of ligand-dependent, specific interactions, surface area	PDBbind v2007	2013
	ΔVinaRF20 [101]	Vina empirical terms, surface area terms	PDBbind v2014 CSAR dataset	2017
XGB	ΔVinaXGB [102]	Vina empirical terms, surface area terms, ligand stability terms, bridge water terms	PDBbind v2016 CSAR dataset	2019
	ΔLinF9XGB [103]	A series of gauss terms characterizing protein–ligand interactions, surface area terms, ligand descriptors, bridge water terms and pocket features	PDBbind CSAR dataset BindingDB	2022
ERT	ET-score [104]	Distance-weighted interatomic contacts between protein and ligand	PDBbind v2016	2021
GBT	AGL-Score [105]	Algebraic graph theory-based features of protein–ligand complex	PDBbind	2019
	ECIF-GBT [106]	Protein–ligand atom-type pair counts considering each atom connectivity	PDBbind v2016	2021
NN	NNScore 1.0 [107]	Descriptors of specific interactions and ligand-dependent	MOAD PDBbind	2010
	NNScore 2.0 [108]	Vina empirical terms, protein–ligand atom-type pair counts in predefined distance cutoff	MOAD PDBbind	2011
CNN	AtomNet [109]	Local structure-based 3D grid from protein–ligand structures	DUD-E	2017
	Pafnucy [110]	Atom property-based 3D grid from protein–ligand structures	PDBbind v2016	2017
	Kdeep [111]	Atom type-based 3D grid from protein–ligand structures	PDBbind v2016	2018
	OnionNet [112]	Rotation-free element-pair specific contacts between protein and ligand atoms in different distance ranges	PDBbind v2016	2019
GNN	PotentialNet [113]	Atom node feature and distance matrix	PDBbind v2007	2018
	graphDelta [114]	Atom node features considering local environment and distance matrix	PDBbind v2018	2020
	SIGN [115]	Distance matrix of atom nodes and angle matrix of bond edges	PDBbind v2016	2021