基于多维伪F统计量的基因序列图形动态聚类研究  

ON DYNAMIC CLUSTERING METHOD FOR GENE SEQUENCE PATTERN BASED ON MULTIDIMENSIONAL PSEUDO F-STATISTICS

作　　者：曹晓霞[1] 赵亮[2] 吴力争[2] 

机构地区：[1]丽水学院计算机与信息工程学院,浙江丽水323000 [2]大连理工大学电子信息工程学院,辽宁大连116024

出　　处：《计算机应用与软件》2009年第9期83-85,98,共4页Computer Applications and Software

基　　金：辽宁省高新技术专业化人才培养重点计基金(200701A)

摘　　要：利用动态调整聚类个数的思想,在模糊C-均值聚类算法基础上引入基于多维PFS判别函数,提出一种基于多维伪F统计量的基因表达动态C-均值聚类算法。以H5N1病毒基因序列数字特征提取为例,在聚类分析过程中直接利用数字特征矩阵作为分析数据,结果表明该算法可以动态调整聚类个数,给出最佳聚类数目,从而获得较好的聚类质量。The idea of dynamic adjustment for cluster count has been made use of in this paper, and a new dynamic C-means clustering algorithm for Genes expressed data has been proposed based on multi-dimension pseudo F-statistics. Put the numerical character of H5N1 virus gene sequence abstracted as an example,in the procedure of clustering analysis the numerical characterization matrix is directly used as analytical data source, the experiment results show that the algorithm can adjust cluster number and gain a prime number of clustering, which thus argues that this algorithm can attain better clustering quality.

关 键 词：模糊C-均值聚类 伪F-统计量 基因表达 基因序列 特征提取 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] O159[自动化与计算机技术—计算机科学与技术]