基于改进LF算法的PPI网络聚类方法

PPI Network Clustering Method Based on Improved LF Algorithm

研究蛋白质相互作用(Protein-Protein Interaction,PPI)网络是理解生命活动的重要途径之一,运用数据挖掘中的聚类分析方法去研究蛋白质相互作用已经成为生物信息学的热门领域.作为一种在PPI网络聚类分析中新兴的智能优化算法,蚁群聚类算法因其固有的简单性、灵活性和鲁棒性显示出了巨大应用潜力.将蚁群聚类算法应用到PPI网络聚类中,并进行了改进,提出了一种新的PPI网络聚类算法.算法引入边关联强度的概念并以边关联强度作为相似度参数,对蚁群聚类算法中的拾起/放下策略加以改进.算法在MIPS数据库中的PPI数据集上进行了聚类测试,实验结果表明新算法的聚类效果和运行效率都较为理想.

Protein-Protein interactions network research is one the most important ways to understand of life activities. Using the clustering analysis method in data mining to study protein interactions has become popular in the field of bioinformatics. In PPI network clustering analysis, ant colony clustering algorithm as a new intelligent optimization algorithm, shows great potential for application because of its inherent simplicity, flexibility and robustness. Ant colony clustering is introduced into PPI network clustering, and is improved. A new PPI network clustering method is presented. This method modifies the pickup/drop rules of Ant colony algorithm by means of introducing the concept of edge intensity, which regards the edge intensity as similarity paramter to cluster the PPI network. Finally the algorithm is tested on the PPI data in MIPS database. The simulation results show that the new algorithm has better clustering effect and running efficiency.