Partial ordering of dissimilarity measures for gene clustering

Clustering methods are widely used in the analysis of microar- ray data for their ability to discover co–regulated genes. In a previous work we introduced, in a hierarchical clustering context, a theoretical framework for the comparison of dissimilarity measures on the basis of their ability to identify functional modules consisting of a transcription factor and the associated target genes. In this paper we extend these results by including in the analysis a set of dissimilarity measures based on the “1 − absolute value of correlation coefficient” proximity between genes. We show that such theoretical framework allows one to obtain a partial ordering of the considered measures in which we identify three minimal elements that are then compared on the basis of both simulated and real data.