Exploiting structural and topological information to improve prediction of RNA-protein binding sites

BMC Bioinformatics

Table 4 Predictive power of combined features on RB144 data set.

AUC	MCC	SN[%]	SP[%]	Classifier	Patch type	Features
0.77	0.30	73.8	67.1	NB	sequential	Profile+aRC
0.78	0.33	84.6	54.8	NB	topological	Profile+aRC+aBC
0.79	0.32	76.9	65.7	NB	spatial	Profile+ASA
0.79	0.34	84.3	56.2	NB	spatial	Profile+ASA+aBC+aRC
0.79	0.34	78.4	66.0	SVM	sequential	Profile+aRC
0.81	0.36	81.1	65.1	SVM	topological	Profile+aRC+aBC
0.82	0.38	81.1	66.7	SVM	spatial	Profile+ASA
0.83	0.39	82.0	66.8	SVM	spatial	Profile+ASA+aBC+aRC

Predictive power of combined features on RB144 data set using five-fold cross-validation tests. C-value for SVM was 1.0 and γ-value of RBF-kernel was 0.01. Cost factor was 5.7.

ISSN: 1471-2105