Functional module detection of protein⁃protein interaction ( PPI) network has been a major challenge i⁃dentified recently by medical researchers. It allows understanding and recognizing the interaction between proteins in an efficient manner. In this study, topological module detection methods, popular in the field of complex protein networks, were applied to the PPI network to obtain these modules, followed by a biological analysis of the topolog⁃ical modules. The interaction mechanism was observed for only 10%~20% of the protein pairs because of incom⁃plete PPI data. Furthermore, the data for noise interaction always existed in PPI;therefore, the number of biologi⁃cally precise modules decreased according to topological community⁃detection methods. In this study, the protein complex data was incorporated into the PPI network to identify more biologically precise protein modules. K⁃Means clustering and non⁃negative matrix factorization algorithms were used to segregate the PPI network into different modules. Gene ontology (GO) and pathway analysis were conducted for each of these modules to quantify their bio⁃logical significance. The results of the experiments showed that the modules detected by combining the protein com⁃plex and PPI network demonstrate a higher tendency to achieve larger homogeneity values compared with those de⁃tected using GO and pathway analysis.%人类蛋白互作网络中功能模块的检测是目前网络医学研究的一个热点问题。好的功能模块可以帮助我们更好地去理解和认识蛋白质相互作用的分子机理。近年来的一些研究大多数是基于复杂网络中的拓扑模块发现算法对蛋白质相互作用网络进行模块划分，然后对其进行生物学上的功能研究。由于PPI网络中的蛋白之间相互作用的数据获取的不完整，相关研究表明目前人类只获得了人类蛋白之间相互作用数据的10％～20％，其中已经获取的数据中还包含着一些噪声，这就导致基于拓扑结构的社团检测算法的精度降低。为了克服这个问题，本文将蛋白质复合体数据融入到模块检测算法中，分别使用K⁃Means和NMF算法对PPI网络进行模块划分，然后从基因本体和通路2个方面对检测到的模块进行功能分析。实验结果表明融合了蛋白质复合体的PPI网络更容易得到具有生物学意义的功能模块。
展开▼
