Co-clustering algorithm for the identification of cancer subtypes from gene expression data

Cancer has been classified as a heterogeneous genetic disease comprising various different subtypes based on gene expression data. Early stages of diagnosis and prognosis for cancer type have become an essential requirement in cancer informatics research because it is helpful for the clinical treatment of patients. Besides this, gene network interaction which is the significant in order to understand the cellular and progressive mechanisms of cancer has been barely considered in current research. Hence, applications of machine learning methods become an important area for researchers to explore in order to categorize cancer genes into high and low risk groups or subtypes. Presently co-clustering is an extensively used data mining technique for analyzing gene expression data. This paper presents an improved network assisted co-clustering for the identification of cancer subtypes (iNCIS) where it combines gene network information with gene expression data to obtain co-clusters. The effectiveness of iNCIS was evaluated on large-scale Breast Cancer (BRCA) and Glioblastoma Multiforme (GBM). This weighted co-clustering approach in iNCIS delivers a distinctive result to integrate gene network into the clustering procedure.