Analysis of Gene Expression Data by Evolutionary Clustering Algorithm

An evolutionary clustering algorithm has been proposed to cluster genes having similar expression profiles. This algorithm is a hybrid of clustering algorithm and evolutionary computation. An evolutionary computation uses an algorithm that utilizes the three biological principles of evolution, namely mutation, crossover, and natural selection to solve an optimization problem. The idea of mutual information is used to find the genetic distances between the genes. It takes into account the similarities as well as positive and negative correlations among the genes. Similar types of genes are clustered together. A large search space of gene expression levels are incorporated using genetic algorithm so that it might lead to better optimization of gene clustering problems. A study on some cancerous microarray gene expression datasets and a comparison with some existing algorithms proves that the as-used algorithm is superior. The algorithm is used to find the top candidate genes responsible for gastric cancer. © 2017 IEEE.