MicNet

MicNet performs microbial network analysis and visualization to characterize co-occurrence patterns, community structure, and stability in amplicon sequencing datasets.

Key Features:

• Visualization Module: Utilizes UMAP (Uniform Manifold Approximation and Projection) for dimensionality reduction and HDBSCAN (Hierarchical Density-Based Spatial Clustering of Applications with Noise) for density-based clustering to visualize high-dimensional microbial datasets.
• Enhanced Network Analysis Module: Integrates an improved SparCC (Sparse Correlations for Compositional data) capable of handling significantly larger datasets and supports construction and analysis of co-occurrence networks using network theory metrics.
• Novel Analytical Approaches: Implements structural balance metrics and methods for uncovering the topology of co-occurrence networks to assess community stability and dynamics.
• Implementation: Implemented in Python.

Scientific Applications:

• Kombucha network: Applied to a kombucha network consisting of 48 Amplicon Sequence Variants (ASVs) to analyze known microbial interactions.
• SparCC validation: The enhanced SparCC module was rigorously tested to demonstrate improved handling and interpretation of large-scale microbial datasets.
• Archean Domes dataset: Applied to a dataset from Archean Domes encompassing over 2,000 ASVs to examine large-scale ecological network structure.

Methodology:

Uses UMAP for dimensionality reduction, HDBSCAN for density-based clustering, an improved SparCC (Sparse Correlations for Compositional data) for correlation estimation, construction and analysis of co-occurrence networks, and computation of structural balance and network topology metrics.