Microarray gene expression data provides a simultaneous gene expression profile of thousands of genes. A handful of genes among the thousands of genes play dominant roles in a disease process. It is impractical to find these faulty genes from experimental data. A computational approach is necessary to find the key genes and hence it is an important area of research in bioinformatics. A new algorithm is developed in this work to identify the candidate genes of a cancer from microarray gene expression data. Gene clustering enables identification of co-expressed genes in specified biological conditions. Many algorithms have inherent limitations to understand the accurate information. This model is a hybrid of clustering algorithm and genetic. A genetic algorithm utilizes the three biological principles of evolution, such as selection, recombination, and mutation to solve an optimization problem. An interdependence measure between the genes is based on a metric which can be mutual information. In many conventional algorithms, a Euclidean genetic distance measure is used. This is basically the difference between gene expression values. Genes are similar if their expressions levels are very close. Similarity of the gene expression levels, and also positive and negative correlations between the genes are taken into account while clustering them. This property is used in mutual information which introduces a parameter called interdependence measure. This says roughly how the two genes are correlated. The genes having higher interdependence measures are the top candidate genes responsible for a diseased state. These top genes are believed to be faulty genes that contain the most diagnostic information for cancer. An analysis is done on gastric cancer, colon cancer, and brain cancer microarray gene expression datasets to pinpoint the faulty genes. In comparison with many existing computational tools, the top candidate genes found by this evolutionary computational model, are able to classify the known samples into cancerous and normal ones with high accuracy. The new model called as evolutionary clustering algorithm also creates more even-like distribution of genes in the clusters. Furthermore, the present computational tool is more coherent in clustering the genes having large differences in their expression values. This information-theoretic computational method can be potentially applied to the big data analysis from other sources.