Table 4 Prediction performance of different classifiers for PLP-interacting residues (PLPIRs)
Feature |     Classifier | SN | SP | ACC | MCC |
|---|---|---|---|---|---|
 Binary | SVM (Threshold = −0.7) | 77.02 ± 0.72 | 83.17 ± 0.27 | 82.62 ± 0.28 | 0.42 ± 0.01 |
SVM (Threshold = −0.5) | 54.76 ± 1.34 | 95.81 ± 0.14 | 92.08 ± 0.18 | 0.51 ± 0.01 | |
BayesNet | 41.76 ± 0.81 | 88.94 ± 0.49 | 84.65 ± 0.40 | 0.26 ± 0.01 | |
ComplementNaiveBayes | 75.82 ± 1.74 | 77.14 ± 0.35 | 77.01 ± 0.23 | 0.34 ± 0.01 | |
NaiveBayes | 52.20 ± 1.50 | 91.18 ± 0.17 | 87.64 ± 0.20 | 0.37 ± 0.01 | |
NaiveBayesMultinomial | 59.25 ± 1.06 | 88.51 ± 0.19 | 85.85 ± 0.19 | 0.38 ± 0.01 | |
IBk | 40.02 ± 1.24 | 96.31 ± 0.20 | 91.19 ± 0.21 | 0.41 ± 0.01 | |
|  | RandomForest | 52.93 ± 1.09 | 80.03 ± 0.71 | 77.56 ± 0.65 | 0.23 ± 0.01 |
 PSSM | SVM (Threshold = −0.7) | 90.20 ± 1.04 | 92.61 ± 0.18 | 92.40 ± 0.13 | 0.67 ± 0.00 |
SVM (Threshold = −0.1) | 79.76 ± 0.92 | 98.62 ± 0.13 | 96.91 ± 0.11 | 0.81 ± 0.01 | |
BayesNet | 77.66 ± 0.83 | 77.71 ± 0.35 | 77.70 ± 0.30 | 0.36 ± 0.01 | |
ComplementNaiveBayes | 76.28 ± 1.46 | 89.09 ± 0.54 | 87.93 ± 0.45 | 0.50 ± 0.01 | |
NaiveBayes | 79.40 ± 0.76 | 80.36 ± 0.35 | 80.28 ± 0.27 | 0.40 ± 0.00 | |
NaiveBayesMultinomial | 43.96 ± 0.67 | 98.16 ± 0.08 | 93.25 ± 0.07 | 0.52 ± 0.01 | |
IBk | 76.10 ± 0.82 | 98.80 ± 0.06 | 96.74 ± 0.08 | 0.79 ± 0.01 | |
|  | RandomForest | 62.27 ± 1.76 | 98.02 ± 0.12 | 94.78 ± 0.20 | 0.66 ± 0.01 |