CAJ | 학술논문

亚细胞位点是蛋白质很重要的功能特征.找到一种有效的、可信度高的预测蛋白质位点的方法是很必要的.提出了一种基于马尔科夫模型的改进预测方法.首先,对于一条给定的蛋白质序列,通过计算在马尔科夫模型下20个氨基酸残基的状态转移矩阵,建立一个420维的特征向量,然后利用支持向量机进行训练和预测,最后夹克刀检验证实了该方法的预测精度与以前的马尔科夫模型相比得到了一定的提高.
아세포위점시단백질흔중요적공능특정.조도일충유효적、가신도고적예측단백질위점적방법시흔필요적.제출료일충기우마이과부모형적개진예측방법.수선,대우일조급정적단백질서렬,통과계산재마이과부모형하20개안기산잔기적상태전이구진,건립일개420유적특정향량,연후이용지지향량궤진행훈련화예측,최후협극도검험증실료해방법적예측정도여이전적마이과부모형상비득도료일정적제고.
Subcellular localization is a key functional characteristic of proteins. An effective and reliable prediction system for protein subcellular localizations is necessary. In the present paper, a method based on an improved Markov chain model is proposed. First,a 420-dimentional feature vector is constructed based on the state transition matrix of 20 residues under a Markov model of protein sequences. Then the support vector machine (SVM) is used to train the model and predict the subcellular location of a given peptide sequence. Finally ,the jackknife cross-validation test on the dataset confirms that our prediction accuracy is improved compared with the original Markov-based method[1].