Table 2. Prediction accuracy of the SVR predictors based on eight different sequence encoding schemes that incorporate various combinations of global and local sequence features.
Sequence encoding scheme | Number of features | Number of support vectors | CC | RMSE | MAE | R square |
PB | 300 | 100134 | 0.64±0.02 | 1.90±0.10 | 1.41±0.08 | 0.456±0.02 |
SC | 30 | 99021 | 0.65±0.03 | 1.86±0.11 | 1.34±0.10 | 0.474±0.03 |
PB+PP | 345 | 100045 | 0.66±0.02 | 1.88±0.11 | 1.39±0.09 | 0.537±0.03 |
PB+SC | 330 | 99662 | 0.69±0.03 | 1.77±0.10 | 1.31±0.08 | 0.539±0.03 |
PB+PP+SC | 375 | 99682 | 0.70±0.03 | 1.76±0.11 | 1.30±0.09 | 0.540±0.03 |
PB+PP+SC+DISO | 405 | 99719 | 0.70±0.03 | 1.75±0.11 | 1.29±0.09 | 0.539±0.03 |
PB+PP+SC+DISO+WL | 407 | 99319 | 0.70±0.03 | 1.75±0.11 | 1.29±0.09 | 0.539±0.03 |
ALL | 435 | 103631 | 0.71±0.03 | 1.74±0.10 | 1.28±0.08 | 0.541±0.03 |
All results were evaluated using 5-fold cross-validation method and expressed as mean±standard deviation.