Table 2.
Comparison of ACP predictors trained and tested with the main data set. Results were obtained from AntiCP2.0 [11] and ACPred [10], except AI4ACP.
| Predictors | Classifier | Accuracy | Sensitivity | Specificity | MCC * |
|---|---|---|---|---|---|
| AntiCP | SVM | 0.506 | 1.000 + | 0.012 | 0.070 |
| iACP | SVM | 0.551 | 0.779 | 0.322 | 0.110 |
| ACPred | SVM | 0.535 | 0.856 | 0.214 | 0.090 |
| PEPred-Suite | ensemble approach | 0.535 | 0.331 | 0.738 | 0.080 |
| ACPred-FL | ensemble approach | 0.448 | 0.671 | 0.225 | −0.120 |
| ACPred-Fuse | RF | 0.689 | 0.692 | 0.686 | 0.380 |
| AntiCP_2.0 | ETree | 0.754 | 0.775 | 0.734 | 0.510 |
| iACP-FSCM | SVM | 0.825 | 0.726 | 0.903 | 0.646 |
| AI4ACP | CNN | 0.718 | 0.802 | 0.633 | 0.442 |
*: Matthews Correlation Coefficient. +: Top two ranked methods for each index are presented using text formats: first in boldface, second with underline.