Table 1.

Comparison of binding site identification methods.

Method			Principle	Example Tool	Available	Applicable Conditions	Advantage	Disadvantage
Ligand-specific method			Interaction with different types of ligands	SXGBsite [47]	FOSS	Require specific ligand types	Accurate prediction of sites for the desired ligand type	Poor performance for non-specific ligand types
General-purpose methods	Sequence-based		Residue conservation	Concavity	FOSS	Only known sequence	Effective identification of sequence-conserved sites	Exclude physicochemical characteristics
	Template-based		Sequence similarity	LIBRA-WA	NA	Known protein with high homology in databases	Acceptable predictive ability for conserved sites	Poor prediction of novel sites
	Structure-based	Geometry-based	Geometric characteristics	Sitemap	FOSS	Require specific geometric features	High prediction rates in large and superficially bound cystic cavities	Do not consider ligand binding energy
		Energy-based	Energy of interactions	FTSite	Free	Require excellent ligand binding energy	Superior performance in predicting ligand binding energy	Exclude geometric features
	Consensus-based		Comprehensive assessment of the above four methods	COACH	FOSS	All feasible	Address inter-method limitations	Time consuming with huge amounts of data

Note: FOSS = free and open-source software; NA = not available.