ILys-Khib: Identify lysine 2-Hydroxyisobutyrylation sites using mRMR feature selection and fuzzy SVM algorithm

Abstract

As a new type of histone mark, lysine 2-Hydroxyisobutyrylation (Khib) is known to affect the association between histone and DNA. The accurate identification of Khib sites is significant for further exploration of the biological functions and molecular mechanisms of Khib. In this study, a novel bioinformatics tool named iLys-Khib is developed to predict Khib sites. Three kinds of effective features, amino acid factors, binary encoding, and the composition of k-spaced amino acid pairs are incorporated to encode Khib sites. And the maximum relevance minimum redundancy feature selection algorithm are adopted to remove the redundant features. Moreover, a fuzzy support vector machine algorithm is proposed to handle the noise problem in Khib sites training dataset. As illustrated by 10-fold cross-validation, the performance of iLys-Khib achieves a satisfactory performance with a Sensitivity of 74.48%, a Specificity of 65.77%, an Accuracy of 70.12% and a Matthew's correlation coefficient of 0.4040. Feature analysis shows that the polarity factor features play significant roles in the prediction of Khib sites. These analysis and prediction results might provide some clues for understanding the molecular mechanisms of Khib. A user-friendly web-server for iLys-Khib is available at http://bioinform.cn/iLys_Khib/.