Table 1.
Details about sequence-derived features.
Fig 1.
The flowchart of GA-based ensemble method.
Table 2.
The average AUC scores of individual feature-based models using different values for λ, evaluated on IMMA2 by 20 independent runs of the 10-CV.
Table 3.
The average performances of different individual feature-based models, evaluated on IMMA2 by 20 independent runs of the 10-CV.
Table 4.
The absolute values of correlation coefficients of AUC scores yielded by individual feature-based models
Table 5.
The average performances of models merging different feature vectors, evaluated by 20 independent runs of the 10-CV.
Table 6.
The average performances of GA-based ensemble method on benchmark datasets, evaluated by 20 runs of 10-CV.
Table 7.
The frequencies of features in the optimal feature subsets.
Table 8.
The average performances of different models evaluated by 20 independent runs of 10-CV.
Table 9.
The statistics of improvements over benchmark methods (significance level 0.05).