CogNet: classification of gene expression data based on ranked active-subnetwork-oriented KEGG pathway enrichment...

Area of Science:

• Bioinformatics
• Computational Biology
• Machine Learning

Background:

• Traditional gene selection methods lack biological relevance, focusing solely on model performance.
• Integrating biological knowledge into gene selection is crucial for building meaningful models.
• Existing computational tools do not sufficiently incorporate external biological resources.

Purpose of the Study:

• To develop a novel computational approach, CogNet, for integrative gene selection.
• To exploit biological knowledge for grouping genes in machine learning tasks.
• To create a biologically relevant model for classification and ranking.

Main Methods:

• CogNet utilizes pathfindR for biological grouping of genes.
• It performs KEGG pathway enrichment analysis on active subnetworks.
• The approach ranks pathways and identifies differentially expressed genes within them.

Main Results:

• CogNet identifies significant KEGG pathways that accurately classify data.
• It provides a list of differentially expressed genes within these pathways.
• CogNet demonstrates competitive performance against existing methods like maTE and SVM-RCE.

Conclusions:

• CogNet effectively integrates biological knowledge for gene selection and classification.
• The tool enhances the interpretability of gene expression data and pathway roles.
• CogNet offers a biologically meaningful approach to gene selection in machine learning.