• 更多栏目

    顾宏

    • 教授     博士生导师   硕士生导师
    • 性别:男
    • 毕业院校:浙江大学
    • 学位:博士
    • 所在单位:控制科学与工程学院
    • 学科:模式识别与智能系统
    • 办公地点:创新园大厦B0715
    • 电子邮箱:guhong@dlut.edu.cn

    访问量:

    开通时间:..

    最后更新时间:..

    iLM-2L: A two-level predictor for identifying protein lysine methylation sites and their methylation degrees by incorporating K-gap amino acid pairs into Chou's general PseAAC

    点击次数:

    论文类型:期刊论文

    发表时间:2015-11-21

    发表刊物:JOURNAL OF THEORETICAL BIOLOGY

    收录刊物:PubMed、SCIE、Scopus

    卷号:385

    期号:385

    页面范围:50-57

    ISSN号:0022-5193

    关键字:K-spaced amino acid pair; Multi-label classification; Post-translational modification; Support vector machine

    摘要:As one of the most critical post-translational modifications, lysine methylation plays a key role in regulating various protein functions. In order to understand the molecular mechanism of lysine methylation, it is important to identify lysine methylation sites and their methylation degrees accurately. As the traditional experimental methods are time-consuming and labor-intensive, several computational methods have been developed for the identification of methylation sites. However, the prediction accuracy of existing computational methods is still unsatisfactory. Moreover, they are only focused on predicting whether a query lysine residue is a methylation site, without considering its methylation degrees. In this paper, a novel two-level predictor named iLM-2L is proposed to predict lysine methylation sites and their methylation degrees using composition of k-spaced amino acid pairs feature coding scheme and support vector machine algorithm. The 1st level is to identify whether a query lysine residue is a methylation site, and the 2nd level is to identify which methylation degree(s) the query lysine residue belongs to if it has been predicted as a methyllysine site in the 1st level identification. The iLM-2L achieves a promising performance with a Sensitivity of 76.46%, a Specificity of 91.90%, an Accuracy of 85.31% and a Matthew's correlation coefficient of 69.94% for the 1st level as well as a Precision of 84.81%, an accuracy of 79.35%, a recall of 80.83%, an Absolute_Ture of 73.89% and a Hamming_loss of 15.63% for the 2nd level in jackknife test. As illustrated by independent test, the performance of iLM-2L outperforms other existing lysine methylation site predictors significantly. A matlab software package for iLM-2L can be freely downloaded from https://github.com/juzhe1120/Matlab_Software/blob/master/iLM-2L_Matlab_Softwaresar. (C) 2015 Published by Elsevier Ltd.