Predicting DNA-binding proteins using feature fusion and MSVM-RFE

Abstract

DNA-binding proteins play a vital important role in cell activities. Prediction of DNA-binding proteins is an important but not fairly solved problem. Currently prediction of DNA-binding proteins via calculation method is a research hotspot. In this paper, we adopt MSVM-RFE for feature selection to those high-dimensional features generated in multiclass feature fusion process and obtain a representative feature subset. The feature subset is evaluated by thirty times 10-fold cross-validation test. At last, we verified the effectiveness of this method compared with method DNA-Prot and other method through three typical datasets. The results demonstrate that this method of predicting DNA-binding proteins has better effect.