Binary Political Optimizer for Feature Selection Using Gene Expression Data

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• DNA Microarray technology enables simultaneous gene expression analysis of thousands of genes.
• Selecting relevant genes from large-scale expression data is a critical challenge in bioinformatics.
• Feature selection techniques are essential for improving classification accuracy by reducing data dimensionality.

Purpose of the Study:

• To propose a binary version of the Political Optimizer (PO) algorithm for effective gene expression data feature selection.
• To introduce two transfer functions, Sigmoid (BPO-S) and V-shaped (BPO-V), for the binary PO.
• To evaluate the performance of the proposed binary PO methods on biological datasets.

Main Methods:

• Development of a binary Political Optimizer (BPO) algorithm.
• Implementation of BPO with Sigmoid (BPO-S) and V-shaped (BPO-V) transfer functions.
• Evaluation using 9 biological gene expression datasets and comparison with 8 established binary metaheuristics.

Main Results:

• The proposed BPO methods, particularly BPO-V, showed prevalent performance in feature selection.
• BPO-V outperformed other binary metaheuristics in identifying relevant genes across multiple datasets.
• The binary adaptation of the Political Optimizer effectively addresses the challenge of gene selection in microarray data.

Conclusions:

• Binary Political Optimizer (BPO) is an effective approach for feature selection in gene expression data.
• BPO-V demonstrates superior performance compared to other metaheuristics for this task.
• The proposed methods contribute to improving the accuracy of classification in genomic studies.