INVESTIGATION OF DIFFERENTIALLY EXPRESSED GENE RELATED TO HUNTINGTON'S DISEASE USING GENETIC ALGORITHM

Abstract

neurodegenerative diseases have complex pathological mechanisms. Detecting disease-associated genes with typical differentially expressed gene selection approaches are ineffective. Recent studies have shown that wrappers Evolutionary optimization methods perform well in feature selection for high dimensional data, but they are computationally costly. This paper proposes a simple method based on a genetic algorithm engaged with the Empirical Bays T-statistics test to enhance the disease-associated gene selection process. The proposed method is applied to Affymetrix microarray data from Huntington's disease. 40 disease-associated genes are discovered as biomarkers. Moreover, the proposed approach improved the disease-associated gene prediction process. The classification accuracy for selected genes is calculated using the K nearest neighbor with leave-one-out cross-validation. The accuracy ranges from 93.1 to 100 for 3 different brain regions, suggesting the effectiveness and robustness of selected genes.