Predicting sumoylation sites using support vector machines based on various sequence features, conformational flexibility and disorder

Table 5 Comparison of SUMOsu with other predictors.

Method	Threshold	Acc	Sp	Sn	MCC
SUMOsp2.0	Low	0.83	0.83	0.75	0.30
SUMOhydro		0.91	0.91	0.71	0.41
seeSUMO-RF		0.82	0.83	0.75	0.30
seeSUMO-SVM		0.90	0.91	0.67	0.37
SUMOsu		0.96	0.97	0.67	0.56
SUMOsp2.0	Medium	0.91	0.93	0.63	0.38
SUMOhydro		0.92	0.94	0.67	0.43
seeSUMO-RF		0.88	0.88	0.71	0.35
seeSUMO-SVM		0.93	0.95	0.54	0.40
SUMOsu		0.96	0.97	0.58	0.52
SUMOsp2.0	High	0.95	0.96	0.58	0.47
SUMOhydro		0.93	0.95	0.58	0.42
seeSUMO-RF		0.89	0.90	0.67	0.36
seeSUMO-SVM		0.95	0.98	0.38	0.39
SUMOsu		0.96	0.98	0.58	0.54
Regular Expressions	N/A	0.95	0.96	0.71	0.56

The values of accuracy (Acc), specificity (Sp), sensitivity (Sn), and Matthew's correlation efficients (MCC) are obtained from Chen et al. [23] as the exact same independent dataset was employed in this study. Thresholds for SUMOsu was set as -0.5, 0, and 0.5 for low, medium and high, respectively.

ISSN: 1471-2164