随机走步软聚类识别蛋白质网交叠功能模块

认定蛋白质交互网中的功能模块对于理解细胞的组织结构，处理过程及其功能都是至关重要的。由于一个蛋白质可能具有多重功能因此蛋白质功能模块可能会相互交叠。然而，目前已有的一些识别功能模块的软聚类算法有如下两个缺陷：一是一些算法需要阈值来截断聚类；二是有些算法需要预设聚类数目。为了克服这两个缺陷，提出了随机走步软聚类识别算法来认定蛋白质交互网中的交叠功能模块。实验结果表明该算法在无需任何参数的情况下能发现一些重要的交叠功能模块。

Random walk soft clustering method for identifying overlapping functional modules in protein interaction networks

Identification of functional modules in protein interaction networks is crucial to understanding principle of cellular organization,processes and function.As a protein can play different roles,functional modules overlap with each other.However, disadvantages of some overlapping clustering algorithms for identifying functional modules arise from the facts that the clustering number and a threshold for truncating cluster are needed to be assigned previously.A random walk soft clustering method is developed to mine overlapping functional modules in protein interaction networks.The numerical results show that this algorithm can detect many overlapping functional modules without adding any parameters.