EMMIXgene

Provides unsupervised selection and clustering of microarray data using mixture models. Following the methods described in McLachlan, Bean and Peel (2002) doi:10.1093/bioinformatics/18.3.413 a subset of genes are selected based one the likelihood ratio statistic for the test of one versus two components when fitting mixtures of t-distributions to the expression data for each gene. The dimensionality of this gene subset is further reduced through the use of mixtures of factor analyzers, allowing the tissue samples to be clustered by fitting mixtures of normal distributions.