基于进化保守性的蛋白质相互作用位点预测

摘要

To extract experience knowledge from PDB and prediction of protein interaction sites, a design method of SVM is presented.Using protein-sequence profile as the feature vectors to predict protein interaction sites. Starting from the protein sequence level, to sequence spectral sequence of neighboring residues as input feature vectors to predict protein interaction sites. Prediction accuracy of 70.47％, CC = 0.19 1 9. The results show that the use of protein sequence spectrum, combined with SVM algorithm for protein-protein interaction site prediction method is effective.%为了从蛋白质结构数据库中提取经验知识,进行蛋白质作用位点预测,提出了以蛋白质序列谱作为特征向量,采用支持向量机算法进行训练和预测蛋白质相互作用位点的方法.从蛋白质一级序列出发,以序列上邻近残基的序列谱为输入特征向量,采用支持向量机方法构建预测器,来预测蛋白质相互作用位点,预测精度达到70.47%,相关系数CC=0.1919.实验结果表明,利用蛋白质序列谱,结合支持向量机算法进行蛋白质相互作用位点预测的方法是有效的.