Predicting lysine lipoylation sites using bi-profile bayes feature extraction and fuzzy support vector machine algorithm

• Published in: Analytical biochemistry Vol. 561-562; pp. 11 - 17
• Main Authors: Ju, Zhe, Wang, Shi-Yun
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 15.11.2018
• Subjects: Bayes Theorem; bioinformatics; computational methodology; Feature extraction; Fuzzy Logic; Fuzzy support vector machine; Lipoylation; lysine; Lysine - metabolism; metabolic diseases; palmitoylation; Post-translational modification; prediction; Support Vector Machine; support vector machines; Post-translational modification; ACC; AAC; BE; Lipoylation; AAF; PTM; SVM; MCC; Fuzzy support vector machine; BPB; Feature extraction; CKSAAP; Sn; Sp
• ISSN: 0003-2697; 1096-0309; 1096-0309
• DOI: 10.1016/j.ab.2018.09.007

Lipoylation is a highly conserved post-translational modification which has been found to be involved in many biological processes and closely associated with various metabolic diseases. The accurate identification of lipoylation sites is necessary to elucidate the underlying molecular mechanisms of lipoylation. As the traditional experimental methods are time consuming and expensive, it is desired to develop computational methods to predict lipoylation sites. In this study, a novel predictor named LipoPred is proposed to predict lysine lipoylation sites. On the one hand, an effective feature extraction method, bi-profile bayes encoding, is employed to encode lipoylation sites. On the other hand, a fuzzy support vector machine algorithm is proposed to solve the class imbalance and noise problem in the prediction of lipoylation sites. As illustrated by 10-fold cross-validation, LipoPred achieves an excellent performance with a Matthew's correlation coefficient of 0.9930. Therefore, LipoPred can be a useful bioinformatics tool for the prediction of lipoylation sites. Feature analysis shows that some residues around lipoylation sites may play an important role in the prediction. The results of analysis and prediction could offer useful information for elucidating the molecular mechanisms of lipoylation. A user-friendly web-server for LipoPred is established at 123.206.31.171/LipoPred/. [Display omitted] •A novel predictor is develop to predict lipoylation sites.•The bi-profile bayes encoding is used to predict and analyze lipoylation sites.•The fuzzy SVM is adopted as classifier.•A free online service is available for prediction.