Co-ABC: Correlation artificial bee colony algorithm for biomarker gene discovery using gene expression profile.

Area of Science:

• Bioinformatics
• Computational Biology
• Machine Learning in Genomics

Background:

• Gene expression profiles are high-dimensional and complex.
• Identifying relevant genes is crucial for accurate disease classification and biomarker discovery.
• Existing feature selection methods may struggle with noise and redundancy in genomic data.

Purpose of the Study:

• To propose a novel hybrid method (Co-ABC) for efficient gene selection from gene expression data.
• To enhance the accuracy of cancer classification by identifying a minimal set of informative genes.
• To validate the effectiveness of Co-ABC against existing methods for biomarker discovery.

Main Methods:

• A three-stage hybrid approach: Correlation-based Feature Selection (CFS) for initial filtering, Artificial Bee Colony (ABC) algorithm for informative gene selection, and Support Vector Machine (SVM) for classification.
• Evaluation on six diverse cancer gene expression datasets (binary and multi-class).
• Comparison with re-implemented Co-GA and Co-PSO methods.

Main Results:

• The Co-ABC algorithm achieved accurate classification performance.
• It successfully identified a small subset of predictive genes.
• Experimental results demonstrated superior or competitive performance compared to Co-GA and Co-PSO.

Conclusions:

• Co-ABC is an efficient and effective method for biomarker gene discovery in cancer.
• The hybrid approach successfully handles high-dimensional gene expression data.
• The method facilitates accurate classification with a reduced number of genes, highlighting its potential for clinical applications.