An Intelligent Optimized ReliefF Model for Autism Gene Selection from Microarray Data

Data science is an interdisciplinary domain that has vast applications. One among them is the bioinformatics domain. Bioinformatics is an essential field of science that incorporates computational, statistical and mathematical aspects for various experimental studies. Microarray experiments are vital in identifying the genetic cause of acute diseases. But the massive volume of data generated from microarray experiments makes analysis tedious and time-consuming for researchers. Hence, an efficient method for selecting relevant gene features is essential. This article proposes a new model for autism gene subset selection termed optimized reliefF (OReliefF). The proposed OReliefF employs a new fuzzy multi-verse optimizer (FMVO) to optimize the selection of neighbours. The filter-based feature selection model obtains the gene subset with high classification accuracy and a low error rate. Benchmark datasets of National Center for Biotechnology Information, USA with accession numbers GSE25507, GSE26415 and GSE2704 are used for implementation. The experimental findings show that the proposed autism gene selection model outperformed the state-of-the-art techniques.