Rigorous assessment and integration of the sequence and structure based features to predict hot spots

BMC Bioinformatics

Table 3 The results of different models on 10-fold cross-validation test and independent test

Features	Testing	P	R	F1	AUC
ASA	10-fold	0.58	0.65	0.61	0.62
	Test set	0.60	0.61	0.61	0.57
BC	10-fold	0.66	0.33	0.43	0.67
	Test set	0.71	0.40	0.51	0.58
Phy	10-fold	0.63	0.51	0.55	0.67
	Test set	0.59	0.53	0.56	0.58
ECS	10-fold	0.58	0.27	0.34	0.51
	Test set	0.74	0.18	0.28	0.68
SE	10-fold	0.57	0.53	0.54	0.60
	Test set	0.59	0.61	0.60	0.52
PSSM	10-fold	0.65	0.54	0.58	0.65
	Test set	0.64	0.48	0.55	0.66
Phy+PSSM+ECS+SE	10-fold	0.65	0.65	0.65	0.68
	Test set	0.69	0.68	0.68	0.68
Phy+ASA+BC	10-fold	0.65	0.60	0.61	0.70
	Test set	0.62	0.70	0.66	0.62
Phy+ASA+BC+PSSM+ECS+SE	10-fold	0.66	0.68	0.66	0.72
	Test set	0.65	0.63	0.64	0.66

ASA denotes accessible surface area; BC denotes biochemical contacts; Phy means physicochemical features; ECS denotes evolutionary conservation score; SE means sequence entropy; and PSSM is the abbreviation of Position-Specific Scoring Matrix.

ISSN: 1471-2105