Table 2 Predictive performance of the transfer-learning method and baseline method evaluated based on the five-fold cross validation tests.
From: Knowledge-transfer learning for prediction of matrix metalloprotease substrate-cleavage sites
MMP type | Model* | Accuracy (%) | Sensitivity (%) | Specificity (%) | F-score (%) | MCC |
|---|---|---|---|---|---|---|
MMP-2 | TL | 87.987 | 83.117 | 89.610 | 77.576 | 0.697 |
BL | 85.390 | 72.727 | 89.610 | 71.338 | 0.616 | |
MMP-3 | TL | 88.110 | 82.927 | 89.738 | 77.714 | 0.699 |
BL | 88.110 | 82.927 | 89.837 | 77.714 | 0.699 | |
MMP-7 | TL | 89.474 | 89.474 | 89.474 | 80.952 | 0.744 |
BL | 84.868 | 71.053 | 89.474 | 70.130 | 0.600 | |
MMP-8 | TL | 85.638 | 74.468 | 89.362 | 72.165 | 0.626 |
BL | 85.106 | 72.340 | 89.362 | 70.833 | 0.609 | |
MMP-9 | TL | 84.954 | 70.370 | 89.815 | 70.046 | 0.600 |
BL | 83.333 | 63.889 | 89.815 | 65.714 | 0.548 | |
MMP-12 | TL | 90.263 | 91.053 | 90.000 | 82.381 | 0.764 |
BL | 90.658 | 92.632 | 90.000 | 83.215 | 0.776 |
