基于稳定性的基因表达数据聚类算法选择

聚类是一种常用的基因表达数据处理手段,然而它又是主观的,如何选择符合数据内在分布的聚类算法成为目前急待解决的问题.根据经验,当选择最佳簇数k后,采用合理的聚类算法对目标数据重复聚类时,结果稳定性较好.因此提出一种基于稳定性的聚类算法选择.该方法将聚类结果的簇间分离度、簇内紧致度和聚类结果稳定性三者结合起来.在验证和应用三组数据时发现,比传统的评估方法,基于稳定性的聚类算法选择更客观、更可靠.

A stability-based clustering selection for gene expression data

Clustering algorithm has become a typical tool for gene expression data analysis recently.However,it is subjective.Algo- rithm Selection is an urge problem it facing.Finding that with the correct number of clusters,the results of most suitable clustering al- gorithms are stable while repeated,a stability-based selection method was proposed. It combines compactness of clusters,separation between clusters and stability of clustering results.Three data sets were used to test the ability of this method,and it performed better with a more objective process than traditional methods.