Hybrid black widow optimization with iterated greedy algorithm for gene selection problems.

Area of Science:

• Bioinformatics
• Computational Biology
• Machine Learning

Background:

• Gene Selection (GS) is crucial for reducing complexity in high-dimensional gene expression datasets from DNA Microarrays.
• Microarray data analysis faces challenges due to inherent data imbalance and discrepancies between sample and gene numbers.
• Existing GS methods can be computationally intensive and may not effectively handle data complexities.

Purpose of the Study:

• To develop a simplified and computationally effective approach for attribute selection in microarray gene expression data.
• To enhance the efficiency of gene selection by improving the local search capabilities of optimization algorithms.
• To validate the performance of a novel hybridized algorithm against existing methods.

Main Methods:

• Utilized the Black Widow Optimization algorithm (BWO) for Gene Selection.
• Developed and tested two BWO methodologies: an unaltered BWO and a hybridized BWO with the Iterated Greedy algorithm (BWO-IG).
• Empirically validated the BWO-IG approach using nine benchmark microarray gene expression datasets and compared it against five modern wrapper Feature Selection (FS) methods.

Main Results:

• The hybridized BWO-IG technique demonstrated superior performance over the traditional BWO algorithm.
• BWO-IG showed enhanced local search efficiency, leading to more reliable identification of relevant genes.
• Comparative analysis revealed BWO-IG's superiority in reducing the number of selected genes while achieving high classification accuracy (average 94.426%).

Conclusions:

• The BWO-IG algorithm offers a computationally effective and reliable solution for gene selection in microarray data analysis.
• Hybridization of BWO with Iterated Greedy significantly improves gene pruning and classification accuracy.
• BWO-IG outperforms existing wrapper FS methods, making it a promising tool for bioinformatics research.