Table 6.

List of ML methods to predict the binding score of protein–ligand binding pose

SN	Approach	Technique involved	Feature	Database used	Year
1	Ashtawy et al.’s method [152]	MLR + MARS + KNN + SVM + RF + BRT	Employed ML approaches utilizing physicochemical and geometrical features characterizing protein–ligand complexes	PDBbind	2015
2	Grudininet et al.’s method [153]	Regression	Predicted binding poses and affinities with a statistical parameter estimation	PDBBind + HSP90 dataset + MAP4K dataset	2016
3	Ragoza et al.’s method [154]	Convolutional neural network	Trained CNN scoring function to discriminate binding poses using the differentiable atomic grid format as input	PDBbind	2017
4	Ragoza et al.’s method [155]	Convolutional neural network	Trained and optimized CNN scoring functions to discriminate between correct and incorrect binding poses	CSAR	2017
5	Nguyen1 et al.’s method [156]	Random forest + convolutional neural networks	Used mathematical deep learning for pose and binding affinity prediction	PDBbind	2018
6	Jose et al.’s method [157]	Reinforcement learning	An approach to represent the protein–ligand complex using graph CNN that would help utilize both atomic and spatial features to score protein–ligand poses	PDBbind + self-curated	2021