Tabu search and binary particle swarm optimization for feature selection using microarray data.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene expression profiles offer molecular insights into cellular states, crucial for medical diagnostics.
• Cancer type classification faces challenges due to small sample sizes and a high number of genes in training datasets.
• Effective gene selection is vital for improving predictive accuracy and interpretability in cancer research.

Purpose of the Study:

• To develop an optimized gene selection method for cancer type classification.
• To address the limitations of small sample sizes in high-dimensional gene expression data.
• To enhance classification accuracy and reduce feature dimensionality.

Main Methods:

• A hybrid feature selection approach combining tabu search (TS) and binary particle swarm optimization (BPSO).
• BPSO functions as a local optimizer within each TS generation.
• K-nearest neighbor (KNN) with leave-one-out cross-validation and support vector machine (SVM) with one-versus-rest were used as evaluators.

Main Results:

• The proposed TS-BPSO method effectively simplifies features in gene expression datasets.
• The method achieved higher classification accuracy or utilized fewer features compared to existing methods across 11 benchmark problems.
• Demonstrated significant improvements in predictive performance and feature reduction.

Conclusions:

• The integrated TS-BPSO approach is a powerful tool for gene selection in cancer classification.
• This method offers a robust solution for handling high-dimensional, low-sample-size data in bioinformatics.
• The findings suggest potential for improved diagnostic tools based on optimized gene expression analysis.