Mutational Slime Mould Algorithm for Gene Selection.

Area of Science:

• Bioinformatics
• Computational Biology
• Machine Learning

Background:

• High-dimensional genetic data is prevalent in modern medicine and biology.
• Processing this data presents challenges in computational complexity and dimensionality.
• Effective gene selection is crucial for data-driven decision-making and accurate classification.

Purpose of the Study:

• To propose a novel wrapper gene selection method using an enhanced slime mould algorithm (SMA).
• To address the challenges of feature selection in high-dimensional genetic datasets.
• To improve classification accuracy and reduce computational costs by identifying representative genes.

Main Methods:

• An improved slime mould algorithm (SMA) incorporating Cauchy mutation and differential evolution (DE) crossover strategies was developed.
• A transfer function was used to convert the continuous SMA into a binary version (BISMA) for gene selection.
• The continuous version (ISMA) was validated on 33 classical optimization problems, and the discrete version (BISMA) was tested on 14 gene expression datasets.

Main Results:

• The improved SMA (ISMA) demonstrated a balance between local exploitation and global search capabilities in continuous optimization.
• The binary version (BISMA) achieved the highest accuracy in gene selection across multiple datasets.
• BISMA effectively reduced the number of selected genes while enhancing classification performance compared to existing methods.

Conclusions:

• The proposed enhanced SMA provides an effective approach for wrapper gene selection in high-dimensional genetic data.
• The method offers a robust solution for reducing data dimensionality and improving classification accuracy in bioinformatics.
• This algorithm holds significant potential for applications in clinical practice and biological research requiring precise gene identification.