Table 1 Prediction performance of different classifiers for vitamin-interacting residues (VIRs)
Feature |     Classifier | SN | SP | ACC | MCC |
|---|---|---|---|---|---|
 Binary | SVM (Threshold = −0.8) | 68.57 ± 0.60 | 64.88 ± 0.18 | 65.22 ± 0.21 | 0.20 ± 0.00 |
SVM (Threshold = −0.5) | 29.53 ± 0.83 | 94.71 ± 0.16 | 88.78 ± 0.15 | 0.27 ± 0.01 | |
BayesNet | 54.76 ± 1.44 | 69.64 ± 0.99 | 68.29 ± 0.85 | 0.15 ± 0.01 | |
ComplementNaiveBayes | 67.57 ± 0.90 | 65.16 ± 0.29 | 65.38 ± 0.33 | 0.19 ± 0.01 | |
NaiveBayes | 35.65 ± 0.85 | 89.52 ± 0.22 | 84.62 ± 0.18 | 0.22 ± 0.01 | |
NaiveBayesMultinomial | 40.08 ± 1.04 | 87.67 ± 0.24 | 83.35 ± 0.24 | 0.22 ± 0.01 | |
IBk | 26.67 ± 0.76 | 93.83 ± 0.11 | 87.73 ± 0.15 | 0.22 ± 0.01 | |
|  | RandomForest | 35.48 ± 0.78 | 79.13 ± 0.36 | 75.17 ± 0.31 | 0.10 ± 0.01 |
 PSSM | SVM (Threshold = −0.8) | 78.52 ± 0.64 | 78.61 ± 0.34 | 78.60 ± 0.32 | 0.37 ± 0.01 |
SVM (Threshold = −0.1) | 52.19 ± 1.01 | 96.79 ± 0.03 | 92.73 ± 0.11 | 0.53 ± 0.01 | |
BayesNet | 67.41 ± 0.24 | 64.20 ± 0.06 | 64.49 ± 0.05 | 0.19 ± 0.00 | |
ComplementNaiveBayes | 61.21 ± 0.58 | 78.06 ± 0.23 | 76.53 ± 0.19 | 0.26 ± 0.00 | |
NaiveBayes | 67.64 ± 0.37 | 65.48 ± 0.11 | 65.68 ± 0.09 | 0.20 ± 0.00 | |
NaiveBayesMultinomial | 54.91 ± 0.94 | 83.52 ± 0.21 | 80.92 ± 0.16 | 0.28 ± 0.01 | |
IBk | 50.70 ± 0.90 | 96.91 ± 0.06 | 92.71 ± 0.08 | 0.52 ± 0.01 | |
|  | RandomForest | 61.54 ± 0.64 | 81.52 ± 0.12 | 79.70 ± 0.11 | 0.30 ± 0.01 |