利用基于Normal矩阵的谱平分法挖掘酵母蛋白质相互作用网络中的社团  

Detecting Communities in the Protein-protein Interaction Network of Saccharomyces cerevisiae Using the Spectral Bisection Method Based on Normal Matrix

作　　者：董蕴源[1] 王正华[1] 王勇献[1] 

机构地区：[1]国防科技大学并行与分布处理国家重点实验室,湖南长沙410073

出　　处：《激光生物学报》2008年第1期13-18,共6页Acta Laser Biology Sinica

基　　金：国家自然科学基金项目(60603054)

摘　　要：蛋白质-蛋白质相互作用(protein-protein interaction,PPI)网络是生物网络的重要组成部分,也是后基因组时代的热点研究问题。揭示PPI网络中的社团结构,对于理解其复杂相互作用的结构和动态特征,了解活体细胞的结构和功能都有很大作用。但目前对于PPI网络结构的分析带有很强的试探性,还没有成熟可靠的方法。传统的谱平分法需要预先知道社团的个数,为了克服这一缺点,在无向无权的PPI网络中使用改进后的基于Normal矩阵的谱平分法,得到了55个有生物学意义的社团。实验结果表明:尽管PPI网络中的社团结构不是很明显,基于Normal矩阵的谱平分法依然可以有效地挖掘出其中具有生物学意义的社团结构。Protein-protein interaction network is the most important part of biology network and it is also the most challenging problem of the post-genomic era. Detecting community structure in protein-protein interaction network can help to understand its complex interaction structure and dynamic characteristics; it also can help to comprehend the structure and function of organism cell. But now the analysis of protein-protein interaction network is heuristic and has not efficient way. Previous spectral analysis approaches have the disadvantage that the number of communities needs to be know beforehand. To overcome this, we improved the spectral bisection method based on normal matrix in undirected and no weight network, and then get fifty five communities with biology sense. The experiment results showed that the spectral bisection method based on normal matrix can mine the structure of communities in PPI network effectively, though there is no clear partitioning indication.

关 键 词：PPI网络 社团结构 谱平分法 

分 类 号：Q811[生物学—生物工程] TP391[自动化与计算机技术—计算机应用技术]