Table 2 Performance of SVM classifiers trained with RBF kernel and features based on Amino Acid Composition, PSI-BLAST, PSSM-Profiles and Cascade SVM.
From: VirulentPred: a SVM based prediction method for virulent proteins in bacterial pathogens
Classifier/training features | Sensitivity (%) | Specificity (%) | Accuracy (%) | MCC |
|---|---|---|---|---|
AAC (A) | 70.0 | 74.1 | 72.1 | 0.44 |
Dipeptide Composition(i+1 st ) (B) | 70.0 | 72.3 | 71.1 | 0.42 |
Dipeptide Composition (i+2 nd ) (C) | 70.2 | 73.7 | 72.0 | 0.44 |
Hybrid1 (A+B+C) | 72.6 | 75.1 | 73.9 | 0.48 |
PSI-BLAST search (D) | 52.5 | 51.7 | 52.1 | ---- |
Hybrid2 (A+B+C+D) | 79.2 | 78.8 | 79.0 | 0.58 |
PSSM (E) | 78.1 | 78.1 | 78.1 | 0.56 |
Hybrid3 (A+B+C+D+E) | 79.0 | 80.1 | 79.6 | 0.59 |
Cascade (A+B+C+D+E) | 82.0 | 81.5 | 81.8 | 0.64 |