Identifying communities from multiplex biological networks.

Area of Science:

• Systems biology
• Network science
• Bioinformatics

Background:

• Biological networks integrate diverse gene/protein relationships (e.g., co-expression, physical interactions).
• Traditional methods often aggregate these distinct interaction types into a single network, losing valuable information.
• Multiplex networks, using separate layers for different interaction types, offer a more informative framework.

Purpose of the Study:

• To evaluate community detection methods for multiplex biological networks.
• To compare aggregation, consensus, and multiplex-modularity approaches.
• To assess the performance of these methods with incomplete or heterogeneous network data.

Main Methods:

• Simulated random networks to test community detection algorithms.
• Applied aggregation, consensus, and multiplex-modularity methods.
• Analyzed a 4-layer multiplex biological network of physical and functional interactions.

Main Results:

• Multiplex-modularity significantly outperformed aggregation and consensus methods, especially with incomplete or heterogeneous network layers.
• Community detection in a 4-layer biological network yielded more accurate functional modules compared to aggregated approaches.
• The study demonstrates the advantage of preserving distinct interaction types in separate network layers.

Conclusions:

• Multiplex network analysis provides a superior framework for biological network analysis and community detection.
• Accounting for network multiplexity leads to more precise identification of functional modules.
• The findings highlight the importance of distinguishing interaction types for robust biological network interpretation.