A modified binary particle swarm optimization for selecting the small subset of informative genes from gene...

Area of Science:

• Bioinformatics
• Computational Biology
• Machine Learning

Background:

• Gene expression data is crucial for developing accurate cancer diagnosis and classification systems.
• Selecting informative genes from high-dimensional data with limited samples presents significant computational challenges.
• Existing computational intelligence methods struggle with irrelevant and noisy genes in cancer gene expression datasets.

Purpose of the Study:

• To propose a modified binary particle swarm optimization (BPSO) algorithm for effective selection of informative genes.
• To enhance cancer classification accuracy by identifying a small subset of relevant genes.
• To improve the efficiency and reduce computational time in gene selection processes.

Main Methods:

• Development of a modified binary particle swarm optimization (BPSO) algorithm.
• Introduction of particle speed and a novel position update rule within the BPSO framework.
• Experimental validation using ten diverse gene expression datasets for cancer classification.

Main Results:

• The proposed modified BPSO demonstrated superior performance compared to conventional BPSO and other related methods.
• Achieved higher classification accuracy and selected a more concise subset of informative genes.
• Exhibited reduced computational running times, indicating increased efficiency.

Conclusions:

• The modified BPSO algorithm is highly effective for selecting informative genes in cancer classification.
• This approach addresses challenges associated with high-dimensional gene expression data.
• The method offers a promising computational tool for advancing cancer diagnosis and research.