Fuzzy cluster analysis of bioinformatics data composed of microarray expression data and gene ontology annotations

This paper presents the framework and results of the cluster analysis of a selected set of Arabidopsis (a leafy plant) genes in the presence of insect-feeding and wounding stress. We outline the methodology by which we coupled the results of a microarray experiment with the Gene Ontology (GO) annotations of each gene to produce aggregate relational data. Our method combines two relational matrices: one matrix is derived from a fuzzy GO similarity measure and another is derived from the microarray data using a statistical similarity measure. Finally, we used a fuzzy clustering algorithm (NERFcM) and a validity measure (CCV) to cluster and validate the resulting relational data. Results are presented that outline the functional summarization of the clusters. The methods presented here give microarray researchers additional tools to investigate relations between gene expression and gene functions.