摘要
磷酸化是蛋白质重要的翻译后修饰之一,磷酸化位点的理论识别是计算生物学的重要研究内容。磷酸化位点附近存在保守残基片段,而这种保守性又与激酶类型相关。选择注释数据相对较多的CK2,PKA和PKC三种激酶催化的磷酸化位点作为研究对象,以序列组分特征,残基位置特异性特征和残基的非近邻关联特征为参数,采用延森-香农离散量(Jensen-Shannon Divergence,JSD)作为各特征差异度量,再使用二次判别分析算法组合不同特征,对磷酸化位点进行预测。对CK2,PKA和PKC三种激酶磷酸化位点7-fold交叉检验,总精度分别达到了90%,90%和86%,这一结果要好于当前其它预测模型。
Protein phosphorylation is one of the most important reversible post-translational modifications（PTMs）,and the theoretical recognition of the phosphorylation site is an important task in computational biology researches.The conservative characteristic of residue fragments around the phosphorylation site has some relation with protein kinase.In this paper,CK2,PKA and PKC phosphorylation sites,which have relatively more enzyme annotation data,are chosen as the object of study,and the sequence component characteristic,the characteristic of residue position specificity,as well as the characteristic of residue pairnon-close neighbor relevancy,are taken as parameters.And,Jensen-Shannon Divergence with Quadratic Discriminant analysis（JSDQD）is used as the method for predicting the phosphorylation sites.The 7-fold cross-validation test accuracies of CK2,PKA and PKC are,respectively,90%,90% and 86%,which are higher than those obtained by other prediction models used currently.
出处
《内蒙古工业大学学报（自然科学版）》
2011年第2期108-115,共8页
Journal of Inner Mongolia University of Technology：Natural Science Edition
基金
内蒙古自然科学基金项目(2010BS0104)
关键词
蛋白质磷酸化位点
延森-香农离散量
二次判别分析
protein phosphorylation site
Jensen-Shannon divergence
quadratic discriminant analysis
