Autism Gene Subset Selection from Microarray data – A Wrapper Approach

Abstract

Autism spectrum disorder is a complex neurodevelopment disorder that affects an individual's social behavior. Microarray analysis is an extensively used technique to detect autism. Microarray data can provide additional insight into the etiology of the disorder. Identifying the specific set of genes associated with autism from complex microarray data poses a significant research challenge due to its high dimensionality. However, Gene subset selection is classified as an np-hard problem that can be handled by the meta-heuristic algorithm. In this paper, a novel meta-heuristic Game Theory Based Whale Optimization Algorithm is proposed. The proposed algorithm uses a two-person zero-sum game theory and convergence parameter to increase convergence rate and avoid local optima. The performance of the proposed algorithm is tested with 23 mathematical benchmark functions and compared with other state-of-the-art algorithms. Further, the proposed algorithm is employed as a wrapper-based gene subset selection model with a support vector machine. Furthermore, the outcomes demonstrate that the gene selection model utilizing a wrapper-based approach is capable of effectively identifying a subset of autism-related genes with desirable accuracy.