Method | Performance (%) on Unbalanced training set—1:2 | |||
---|---|---|---|---|
Accuracy | Precision | Recall | Specificity | |
Decision trees | 94.86 ± 0.58 | 95.67 ± 1.29 | 89.48 ± 0.79 | 97.79 ± 0.67 |
Random forest | 98.74 ± 0.15 | 99.10 ± 0.17 | 97.33 ± 0.44 | 99.45 ± 0.09 |
DeepPPI | 97.91 ± 0.38 | 98.37 ± 0.69 | 95.32 ± 1.40 | 99.21 ± 0.34 |
DeepFE-PPI | 98.53 ± 0.20 | 98.41 ± 0.62 | 97.37 ± 0.31 | 99.16 ± 0.31 |
Struct2Graph | 98.91 ± 0.24 | 99.17 ± 0.15 | 97.89 ± 0.17 | 99.52 ± 0.27 |
Method | MCC | F1-score | ROC-AUC | NPV |
---|---|---|---|---|
Decision trees | 88.69 ± 1.28 | 92.47 ± 0.82 | 93.66 ± 0.58 | 94.47 ± 0.50 |
Random forest | 97.25 ± 0.33 | 98.20 ± 0.22 | 99.71 ± 0.08 | 98.56 ± 0.24 |
DeepPPI | 95.29 ± 0.85 | 96.81 ± 0.60 | 99.29 ± 0.25 | 97.70 ± 0.67 |
DeepFE-PPI | 96.76 ± 0.45 | 97.88 ± 0.31 | 99.41 ± 0.17 | 98.59 ± 0.09 |
Struct2Graph | 97.59 ± 0.51 | 98.43 ± 0.30 | 99.73 ± 0.18 | 98.87 ± 0.16 |