Unsupervised gene selection using biological knowledge : application in sample clustering.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene expression data classification is crucial for disease diagnosis and treatment planning.
• High-dimensional and redundant gene expression data present a significant challenge in bioinformatics.
• Gene selection is vital for reducing data complexity and improving classification accuracy.

Purpose of the Study:

• To explore the use of biological knowledge from Gene Ontology for unsupervised gene selection.
• To reduce feature space dimensionality in gene expression data.
• To improve sample classification accuracy by integrating biological insights into gene selection.

Main Methods:

• An unsupervised feature selection technique leveraging Gene Ontology knowledge was developed.
• No class label information was used during the gene selection process.
• A multi-objective clustering approach was applied to samples in the reduced gene space.

Main Results:

• Incorporating biological knowledge into gene selection significantly reduces feature space dimensionality.
• The proposed method enhances the accuracy of sample classification.
• The reduced gene space demonstrated strong biological significance, validated through rigorous tests.

Conclusions:

• Biological knowledge-guided gene selection is effective for dimensionality reduction in gene expression data.
• The developed technique improves sample classification accuracy compared to existing methods.
• Comparative analysis confirms the superiority of this gene selection-based sample clustering approach.