AN ASSOCIATION RULE MINING APPROACH FOR CO-REGULATED SIGNATURE GENES IDENTIFICATION IN CANCER

Abstract

When a normal cell becomes cancerous there will be change in expression of many genes in that cell. Identification of these changes in gene expression in cancer tissue may lead to the development of novel tools for early diagnosis and effective therapeutics. In this paper we present an association rule mining approach to identify the association between the genes that are differentially expressed in cancer tissue compared to normal tissue. We design an association rule mining algorithm GeneExpMiner for gene expression data mining. Serial Analysis of Gene Expression (SAGE) data related to pancreas cancer is used to demonstrate the approach. It is expected that the approach will help in developing better treatment methodologies for cancer and designing low cost microarray chips for diagnosing cancer. The results have been validated in terms of Gene Ontology and the signature genes that we have identified are matching with the published data.