Skip to main content

Table 1 Performance comparison between the predictions performed using the three sets of features that were generated using the hybrid, the filter-based, and the wrapper-based feature selection methods

From: Prediction of beta-turns at over 80% accuracy based on an ensemble of predicted secondary structures and multiple alignments

Feature selection method Qtotal Qpredicted Qobserved MCC
Hybrid method (90 features) 80.9 62.7 55.6 0.47
Filter-based (54 features) 80.7 64.1 49.5 0.44
Wrapper-based (39 features) 79.6 60.2 51.0 0.42
  1. The tests are based on 7 fold cross validation on the BT426 dataset