Scalable module detection for attributed networks with applications to breast cancer

Area of Science:

• Bioinformatics
• Computational Biology
• Network Science

Background:

• Network module detection aims to find densely connected groups of nodes.
• Integrating node attributes with network structure is a significant challenge.
• Existing methods often struggle to bridge structural and attribute data effectively.

Purpose of the Study:

• To propose a novel Weighted Fast Greedy (WFG) algorithm for attribute-based module detection.
• To effectively integrate network structure and attribute information for enhanced community detection.
• To demonstrate the utility of attribute-based module detection in a biological context.

Main Methods:

• Developed the Weighted Fast Greedy (WFG) algorithm.
• Utilized logistic regression to bridge network structure and attribute spaces.
• Applied the WFG algorithm to a breast cancer dataset, integrating protein-protein interaction networks, gene expression data, and survival outcomes.

Main Results:

• The WFG algorithm successfully identified five modules significant for survival in breast cancer.
• These significant modules contained known cancer pathways and biomarkers, including cell cycle, p53 pathway, BRCA1, BRCA2, and AURKB.
• Neither network structure nor gene expression data alone yielded these critical cancer signatures.

Conclusions:

• Attribute-based module detection is crucial for uncovering biologically relevant network patterns.
• The WFG algorithm provides an interpretable and effective method for integrating network structure and attributes.
• This approach enhances the discovery of cancer biomarkers and survival-associated pathways.