Table 1.

10-fold cross-validation of different combinations of features on RSC758

Feature	ACC	SN	SP	MCC	AUC
D + PSSM + SS + SA + PC	0.650	0.540	0.761	0.309	0.705
D + PSSM + SS + PC	0.653	0.529	0.777	0.316	0.705
D + PSSM + SS	0.658	0.516	0.801	0.330	0.700
D + PSSM	0.644	0.503	0.785	0.300	0.691
D	0.639	0.442	0.835	0.301	0.671
SS	0.555	0.770	0.339	0.121	0.559
PSSM	0.575	0.578	0.573	0.150	0.590
SA	0.557	0.552	0.562	0.114	0.554
PCP	0.525	0.611	0.439	0.051	0.542

The results are sorted by AUC value. The feature set in bold was selected as the optimal

D sequential distance to adjacent cysteines, PSSM PSSM profile, SS predicted secondary structure, SA predicted solvent accessibility, PCP physical-chemical property