HybridGWOSPEA2ABC: a novel feature selection algorithm for gene expression data analysis and cancer classification

DNA micro-array technology has a remarkable impact on biological research, particularly in categorizing and diagnosing cancer and studying gene features and functions. With the availability of extensive collections of cancer-related data, there has been an increased focus on developing optimized Machine Learning (ML) techniques for cancer classification through gene pattern analysis and the identification of specific genes for cancer type categorization. The relevant gene selection for diagnosing and treating cancer poses a significant challenge, which requires efficient feature selection methods.

This study introduces a novel hybrid algorithm, for gene selection, integrating the Grey Wolf Optimizer (GWO), Strength Pareto Evolutionary Algorithm 2 (SPEA2), and Artificial Bee Colony (ABC). This combination uses intelligence and evolutionary computation to enhance solution diversity, convergence efficiency, and exploration and exploitation capabilities in high-dimensional gene expression data. The algorithm was compared with five bio-inspired algorithms using five different classifiers on various cancer datasets to validate its effectiveness in feature selection.

The HybridGWOSPEA2ABC algorithm demonstrated superior performance in identifying relevant cancer biomarkers compared to the conventional bio-inspired algorithms. Comparison with the benchmark algorithms has shown the hybrid approach's enhanced capability in addressing the challenges of high-dimensional data and advancing the gene selection problem for cancer classification.

The novel hybridization algorithm enhances performance by maintaining solution diversity, efficiently converging to optimal solutions, and improving the exploration and exploitation of the search space. This study provides a better understanding of relevant genes for cancer classification and promotes effective methodologies for disease detection and classification.