Feature Selection Based on Genetic Algorithm for Classification of Pre-miRNAs

Precursor miRNAs (pre-miRNAs) are usually extracted to obtain quite a lot of global and intrinsic folding features that include some redundant and useless features. Therefore,it is essential to select the most representative feature subset,which contributes to improve the classification efficiency.We propose a novel feature selection method based on genetic algorithm.The information gain of feature and the redundancy among features are considered in this algorithm.Compared with microPred,the total accuracy of classifier miPredGA which is constructed with our selected features is improved nearly 12%.Our selected feature subset also could be used to train the classifier based on ab initio method,which is beneficial to construct efficient classifier used to classify real pre-miRNAs and pseudo hairpin sequences.