MCentridFS: a tool for identifying module biomarkers for multi-phenotypes from high-throughput data.

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• High-throughput data analysis is crucial for identifying biomarkers.
• Existing methods primarily focus on two-class comparisons, limiting their application to multi-phenotype data.
• There is a significant need for effective multi-classification models to analyze complex biological data.

Purpose of the Study:

• To introduce MCentridFS (Multi-class Centroid Feature Selection), a novel approach for identifying network biomarkers.
• To enable systematic classification of multi-phenotypes using high-throughput data.
• To address the limitations of current methods in handling multi-phenotype biological data.

Main Methods:

• MCentridFS formulates multi-classification using network modules as a binary integer linear programming problem.
• The approach is evaluated on datasets for multi-stage HCV-induced dysplasia, hepatocellular carcinoma, and multi-tissue breast cancer.
• Performance is assessed using five-fold cross-validation and comparison with state-of-the-art multi-classification methods.

Main Results:

• MCentridFS demonstrated high classification and cross-validation rates across tested disease datasets.
• The method outperformed existing state-of-the-art multi-classification techniques.
• Functional enrichment analysis confirmed the biological relevance of the identified network modules.

Conclusions:

• MCentridFS is a robust tool for module biomarker detection in multi-phenotype classification problems.
• The approach effectively leverages both large-scale biological data and network information.
• Identified network modules are significantly associated with relevant biological processes and pathways, validating the method's utility.