Table 4.
Method | AUC mean | AUC std | R50 mean | R50 std | R100 mean | R100 std |
LR | 0.9419 | 0.020 | 0.1148 | 0.031 | 0.1684 | 0.031 |
NB | 0.9389 | 0.003 | 0.0964 | 0.031 | 0.1356 | 0.035 |
RF | 0.9427 | 0.009 | 0.0740 | 0.025 | 0.1263 | 0.030 |
SVM | 0.7645 | 0.091 | 0.0455 | 0.028 | 0.0589 | 0.040 |
MFE | 0.9608 | 0.007 | 0.1341 | 0.023 | 0.1759 | 0.027 |
MFE-FM | 0.9384 | 0.018 | 0.1297 | 0.023 | 0.1713 | 0.025 |
Average AUC and partial AUC scores for six classification methods for PPI prediction in human. LR: Logistic regression; NB: Naive Bayes; RF: Random Forest; SVM: Support Vector Machine; MFE: Mixture-of-Feature-Experts; MFE-FM: Mixture-of-Feature-Experts with missing values filled. Average AUC and partial AUC scores are reported and the standard derivations for each score estimation are also listed in the table. MFE scores are highlighted and it again achieves better AUC/R50/R100 scores compared to the other five classifiers.