A three-way clustering approach to cross-species gene regulation analysis

Many different biological data mining methods have been used in gene expression data analysis. A common method is two-way clustering, also called biclustering, which is used to identify the gene groups that behave similarly under a subset of experimental conditions. This paper introduces a novel approach called three-way clustering (TriWClustering) for cross-species gene regulation analysis, to mine coherent clusters named triclusters in three-dimensional (gene-condition-organism) gene expression datasets. The developed method has been applied to three different gene expression data obtained from NCBI's GEO data collection. Biological and statistical significance of the results are evaluated using Gene Ontology term enrichment analysis and Dunn index (DI) metric, respectively. The experimental results indicate that TriWClustering can find significant triclusters and promote a useful tool for cross species gene regulation analysis.