A fuzzy clustering algorithm influenced by validity indices (FCVI) for recognizing the differentially expressed cancer mediating genes

Abstract

Recent progressions in the field of clustering have by this time led to the emergence of fuzzification systems for different medical applications. Furthermore, in the last few decades computational tools have been designed to advance the proficiencies and abilities of physicians/biologists for detecting cancer mediating biomarkers about their patients. In this article we suggest a methodology based on fuzzy clustering technique for identification of some genes which are having correlation in certain cancers. We have demonstrated our methodology using three gene expression data sets viz. Lung, colon and Leukemia belong to both normal and carcinogenic state. As our algorithm works by analyzing the set of clusters, so it is a major challenge to generate optimum number of clusters for the available data sets. We have used cluster validity index Xie-Beni (XB) index and Fukuyama and Sugeno (FS) index to determine finest number of clusters. The effectiveness of the proposed algorithm is supported by biological and statistical validation.