一种采用新的相似性度量方法的共调控基因动态模糊聚类算法

A dynamic fuzzy clustering algorithm for analyzing co-regulated genes based on new similarity measurement

针对基因间共调控关系的特点和现有共调控基因聚类分析方法的不足,提出一种基于广义信息论中二次互信息的广义相似性度量标准QMISM,并利用免疫遗传算法将高维样本映射到二维空间,进而实现动态模糊聚类和聚类结果可视化.对人工合成数据和真实的基因表达数据的实验结果表明,该算法能得到更好的聚类结果.

Based on the quadratic mutual information, this paper presents a new method of similarity measurement, i.e. QMISM. QMISM is developed on the basis of the special properties and disadvantages of existing clustering algorithm of co - mapped regulated genes. The high dimensional samples are into two dimensional spaces by immune genetic algorithm. The algorithm proposed in this pa- per implements a dynamic fuzzy clustering method and improves the clustering results＇ visualization. Additionally, experiments on synthetic dataset and real gene expression dataset show that the algorithm has better clustering effect.