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

Table 3 Prediction performance of self-consistency, 5-fold cross validation and 10-fold cross-validation tests on the training set.

Evaluation Method	Accuracy	Specificity	Sensitivity	MCC
Self Consistency	0.97	0.98	0.76	0.68
5-fold Cross-validation	0.97	0.98	0.73	0.66
10-fold Cross-validation	0.97	0.98	0.73	0.66
Regular Expressions	0.96	0.97	0.72	0.58

An average of 25 repeats are reported for 5-fold and 10-fold cross validation tests. Standard deviations were less than 0.01, so they were not reported. Regular expressions scan is done by searching [IVLMAP]K.[DE] pattern in the sequence window centering the lysine residue.

ISSN: 1471-2164