Gene Selection Using Multi-Objective Optimization for Disease Classification

The expression levels of thousands of genes in two sample cells are compared using micro array technology. It is possible to make key medical diagnoses, treat a specific disease, track the progression of the disease, and predict how the disease will respond to therapy using those sample cells, depending on their diverse sources. We must choose the most valuable genes from among the thousands of genes present in the microarray because not all of them are helpful in categorizing a particular condition. Our method of gene selection involves two stages. Thousands of genes are reduced to a few hundred (N=100) by utilizing a parametric technique in stage 1, and in stage 2, the collected genes chosen in stage 1 are going through a process of multi-objective genetic algorithm (MOGA), employing Pareto GA (Genetic Algorithm). A smaller fraction of the top-ranked genes chosen in stage 2 is taken into consideration classifying the disease to find with the highest accuracy.