论文部分内容阅读
通过整合蛋白质作用网络拓扑结构信息和蛋白质序列信息,对蛋白质作用网络进行聚类,预测药物靶蛋白。分析了网络密度、节点隶属度及蛋白质相对于聚类团的平均相似度,将蛋白质作用网络分割成具有一定生物学特性的子网络,利用维尔克松秩和检验判断聚类团是药物靶点团还是非靶点团。实验结果表明,与仅利用蛋白质作用网络拓扑结构信息的聚类算法相比,该算法预测精度提高17.4%,能够有效预测药物靶蛋白,推测潜在的药物-靶蛋白作用。
By integrating network topology information and protein sequence information of protein action, the protein action network is clustered to predict drug target protein. The network density, node membership degree and the average similarity of proteins with respect to the cluster were analyzed. The protein network was divided into sub-networks with certain biological characteristics. The Wilcoxon rank sum test was used to determine that the cluster was a drug target Mission or non-target groups. The experimental results show that compared with the clustering algorithm that only uses the network topology information of protein interaction, the prediction accuracy of this algorithm is improved by 17.4%, which can effectively predict drug target protein and speculate the potential drug-target protein.