Integration of biological knowledge in the mixture-of-Gaussians analysis of genomic clustering

The analysis of biological data produced by state of the art high throughput technologies like DNA microarrays presents many challenges due both to the domain itself (e.g. high dimensionality) and the technologies themselves (e.g. noisy data). In this paper we advocate the exploitation of existing biological knowledge in order to guide the cluster analysis of gene expression data. To this end we present a biologically inspired probabilistic model and a modified Expectation-Maximization algorithm for the estimation of its parameters. Finally we perform some initial evaluation of the clustering results of the proposed model.