Glycosylation site prediction using ensembles of Support Vector Machine classifiers

BMC Bioinformatics

Table 3 Performance of classifiers trained to predict C-linked glycosylation sites

Performance Measure	SingleSVM	EnsembleSVM	BalancedSVM
Accuracy	0.83	0.83	0.83
MCC	0.63	0.63	0.63
Sensitivity	0.74	0.79	0.77
Specificity	0.80	0.77	0.78
F-Measure	0.77	0.78	0.77
AUC	0.88	0.91	0.89

Results obtained for C-linked glycosylation using single SVM from unbalanced data (singleSVM), ensemble of SVMs (EnsembleSVM), and single SVM from balanced data (BalancedSVM) for the classification threshold θ = 0.5 on the output probability of the classifier. The classifiers are trained on information derived from the target amino acid residue and its sequence neighbors.

ISSN: 1471-2105