Prediction of catalytic residues using Support Vector Machine with selected protein sequence and structural properties

Table 1 Performance of the SMO classifier in the absence of individual residue property in the optimal 7-attribute set in 10-fold cross-validation analysis

Attribute	MCC in the absence of the attribute				Predictive accuracy in the absence of the attribute, %
	DATASET			Average	DATASET			Average
	1 ^st	2 ^nd	3 ^rd		1 ^st	2 ^nd	3 ^rd
1. conservation_score	0.526	0.536	0.483	0.515	76.17	76.52	73.97	75.55
2. AA_identity	0.668	0.679	0.660	0.669	83.40	83.95	82.97	83.44
3. nearest_cleft_distance	0.708	0.746	0.707	0.720	85.35	87.28	85.32	85.98
4. distance_to_3_largest_clefts	0.724	0.757	0.726	0.736	86.13	87.87	86.30	86.77
5. HB_main_chain_protein	0.725	0.746	0.738	0.736	86.13	87.28	86.89	86.77
6. nearest_cleft_rank	0.740	0.746	0.730	0.739	86.91	87.28	86.50	86.90
7. nearest_cleft_SA_area	0.736	0.746	0.738	0.740	86.72	87.28	86.89	86.96
all attributes (24)	0.720	0.722	0.742	0.728	85.94	86.11	87.08	86.38
selected attributes (7)	0.752	0.753	0.742	0.749	87.50	87.67	87.08	87.42