Table 1. Training sequences dataset classification results (%) for the selected neural networks obtained through APDBase or AAindex encoding, after feature selection with one of the internal classifiers, rf, nb, svm, sda and spls.
Where SI is the sensitivity, SP the specificity, PPV the positive predictive value, NPV the negative predictive value and AC the overall accuracy, averaged after 10-fold stratified resampling.
| SI | SP | PPV | NPV | AC | ||
|---|---|---|---|---|---|---|
| APDBase | NN_APD_rf | 82.6 | 85.0 | 82.3 | 85.1 | 83.8 |
| NN_APD_nb | 77.1 | 77.7 | 74.1 | 80.3 | 77.4 | |
| NN_APD_svm | 87.9 | 86.5 | 84.3 | 89.0 | 86.8 | |
| NN_APD_sda | 73.5 | 80.0 | 75.4 | 78.9 | 76.7 | |
| NN_APD_spls | 80.4 | 79.1 | 76.5 | 82.4 | 79.7 | |
| AAindex | NN_AA_rf | 84.0 | 86.4 | 83.9 | 86.4 | 85.1 |
| NN_AA_nb | 87.4 | 70.2 | 71.2 | 86.9 | 78.0 | |
| NN_AA_svm | 80.1 | 76.9 | 75.1 | 82.8 | 78.7 | |
| NN_AA_sda | 86.8 | 95.6 | 94.9 | 90.3 | 91.9 | |
| NN_AA_spls | 81.0 | 83.3 | 80.1 | 84.2 | 81.7 |