【摘要】 由于蛋白质的相互作用是动态变化的,因此使用常规检测方法从静态PPI网络数据中识别蛋白质复合物具有一定的局限性.本文结合时序基因表达数据,提出了一个基于时序功能保持特征和蚁群聚类的复合物检测算法.算法首先根据相邻时刻的子网结构,选出在相邻时刻都具有表达活性的种子节点集合.然后结合复合物的保持特征,构建一组与前一时刻复合物集合具有功能相似性的初始蛋白质簇集合,并利用蚁群聚类的拾起、放下规则,完成对其他蛋白质的聚类,从而形成最终的复合物.实验结果表明使用时序功能保持特征可以提高复合物预测的准确性,与其他方法相比,新算法在精度方面也具有较好的性能.更多还原

【Abstract】 Due to Protein-Protein Interactions(PPI) in cellular organization are dynamic change,there are certain limitation for regular methods to detect protein complexes from static PPI networks.Combining time series gene expression data,this paper presents a new algorithm based on timing function feature and ant colony clustering to identify complexes in dynamic PPI networks.Firstly,based on the sub-network structures in two adjacent moments,the algorithm selects the set of seeds which are active in the two adjacent moments.Then,the algorithm employs the timing function feature in complexes to construct an initial clustering set which have similar functions with complexes detected in the last moment.Finally,the picking and dropping operations in ant colony clustering is utilized to cluster the rest of proteins in current PPI networks.Experiment results demonstrate that using timing function feature can get more reliable and accurate protein complexes.By comparison with some algorithms,our algorithm has better performance on precision.更多还原