Table 1 Performance metrics of different ML algorithms using the combined features ATC + P2G + AAC_PSSM + D-FPSSM when applied to exactly the same dataset (positives and sampled negatives) used by Wei et al. [22]
From: Improved cytokine–receptor interaction prediction by exploiting the negative sample space
SE | SP | ACC | MCC | AUC | g-means | |
|---|---|---|---|---|---|---|
ATC + P2G + AAC_PSSM + D-FPSSM | ||||||
NB | 94.1 | 68.5 | 81.3 | 0.647 | 0.843 | 80.2 |
A1DE | 95.1 | 69.5 | 82.3 | 0.668 | 0.910 | 81.2 |
SMO-RBF | 93.6 | 86.7 | 90.1 | 0.805 | 0.901 | 90.1 |
SMO-PolyK | 92.6 | 77.3 | 85.0 | 0.708 | 0.850 | 84.6 |
SMO-PuK | 90.6 | 88.2 | 89.4 | 0.788 | 0.894 | 89.4 |
IBK | 91.6 | 84.2 | 87.9 | 0.761 | 0.879 | 87.8 |
Bagging | 85.7 | 79.8 | 82.8 | 0.656 | 0.904 | 82.7 |
RF | 89.2 | 81.3 | 85.2 | 0.707 | 0.935 | 85.1 |
Wei et al | 92.6 | 83.3 | 87.9 | 0.762 | – | 87.8 |