Table 3. Performance of SVM based models using dipeptide composition as input feature on whole peptide dataset and nine terminus datasets.
| Approach | Sn(%) | Sp (%) | Acc (%) | MCC |
|---|---|---|---|---|
| Whole peptide | 75.7 | 73.8 | 74.8 | 0.50 |
| NT5 | 66.4 | 66.4 | 66.4 | 0.33 |
| CT5 | 57.1 | 57.9 | 57.6 | 0.15 |
| NTCT5 | 64.5 | 60.8 | 62.6 | 0.25 |
| NT10 | 70.1 | 73.8 | 72.0 | 0.44 |
| CT10 | 63.2 | 70.1 | 66.7 | 0.33 |
| NTCT10 | 65.4 | 74.8 | 70.1 | 0.40 |
| NT15 | 75.3 | 72.8 | 74.1 | 0.48 |
| CT15 | 74.1 | 77.8 | 75.9 | 0.52 |
| NTCT15 | 72.8 | 76.5 | 74.7 | 0.49 |
Sn: Sensitivity; Sp: Specificity; Acc: Accuracy; MCC: Matthew’s Correlation Coefficient.