MicNet toolbox: Visualizing and unraveling a microbial network

Area of Science:

• Microbial Ecology
• Network Science
• Bioinformatics

Background:

• Microbial communities exhibit complex interactions.
• Network theory offers a powerful framework to study these interactions.
• Existing tools may not scale to large datasets or offer advanced analyses.

Purpose of the Study:

• To present an open-source Python toolbox (MicNet) for microbial network analysis.
• To provide enhanced capabilities for co-occurrence network construction and analysis.
• To facilitate the study of microbial community structure and interplay.

Main Methods:

• Utilizes UMAP for dimensionality reduction and HDBSCAN for clustering.
• Includes an enhanced version of SparCC for network inference, supporting larger datasets.
• Applies network theory analyses, including structural balance metrics and topology discovery.

Main Results:

• The toolbox was validated on a kombucha microbial network (48 ASVs).
• Improvements in the enhanced SparCC version were confirmed.
• Demonstrated application on a large Archean Domes dataset (>2,000 ASVs).

Conclusions:

• MicNet provides a user-friendly and powerful tool for microbial network analysis.
• The toolbox supports researchers with varying programming experience.
• Facilitates deeper understanding of microbial community dynamics through network approaches.