Target Clustering of Genes by Normal Mixture Model in Microarray Analysis

To cluster genes in microarray gene expression data experimented over known classes (conditions), this paper provides a model-based gene cluster method. This method takes the normality for component densities from so called splitting and averaging of class data where the number of components is given, and designs component means by general linear model subject to various linear restrictions in order to induce the targeted clusters that an analyst previously want to discover.