Abstract: The purpose of this study is to evaluate a novel clustering technique for clustering and detecting the isolated points in protein-protein interaction network，which iteratively refines clusters based on a combination of the K-means clustering algorithm and the Arithmetic Average Minimum Value（AAMV）.The result is that the algorithm is found to be effective at detecting clusters and identifying the isolated points in the protein-protein interaction network graph with regard to human Alzheimer’s Disease.The algorithm outperforms competing approaches and is capable of effectively predicting the function-unknown protein function.

Key words: protein-protein interaction network graph, Arithmetic Average Minimum Value（AAMV）, K-means clustering method, criterion function

摘要： 依据人类AD（Alzheimer’s Disease）相关蛋白质相互作用网络图，利用基于算术平均最小值——AAMV（Arithmetic Average Minimum Value）的K-means聚类方法对蛋白质进行聚类并预测4个孤立蛋白质的功能。分析结果表明：所得结果与用Maryland Bridge 法及Korbel法所得结果非常相似。

关键词: 蛋白质相互作用网络图, 算术平均最小值, K-means聚类方法, 准则函数