Table 1.

Summary of state-of-the-art non-template-based functional residue prediction approaches

Method	Reference	Description	F-Score
Sarig Server	Amitai et al. (2004)	Structure-based network analysis	0.137
Youn et al.	Youn et al. (2007)	Support vector machine (SVM) using sequence and structural features	0.279
CRPred	Zhang et al. (2008)	SVM using sequence features	0.282
DISCERN	Sankararaman et al. (2010)	Logistic regression model using phylogenomic and structural inputs	0.286
Conservation-distance-aa	Fajardo and Fiser (2013)	Artificial neural network (ANN) using sequence and structural features	0.269
Wong et al.	Wong et al. (2013)	SVM to find ligand-binding pockets based on structure and sequence properties	0.342
LigandRFS	Chen et al. (2014)	Random forest (RF) classifier using sequence features	0.344
CRHunter (non-template- based portion)	Sun et al. (2016)	SVM using sequence and structural features generated by Delaunay triangulation and Laplacian transformation of protein structures	0.350

Note: F-Scores are taken as reported in the original publications or calculated from reported precision and recall values. The exception is Sarig Server, which does not report F-Score in its original publication; the listed F-Score is taken from the ‘Conservation-distance-aa’ publication, where Sarig Server was also benchmarked.