MetaNet

MetaNet constructs microbial co-occurrence networks and detects biologically significant modules from metagenomic count data to infer interactions within microbial communities, implemented in MATLAB.

Key Features:

• Network Construction: Constructs bacterial co-occurrence networks from metagenomic count data by adapting pairwise similarity/distance measures originally used for ecological sample analysis to taxa-level comparisons.
• Sparse Inverse Covariance Approach: Extends the sparse inverse covariance method to estimate a sparse inverse of a similarity matrix derived from count data, enabling inference of direct associations among thousands of bacterial taxa.
• Module Detection: Identifies subnetworks (modules) within constructed networks to highlight biologically significant groups of co-occurring taxa.
• Efficiency with Large Datasets: Optimized for analysis of large-scale metagenomic datasets containing thousands of bacterial taxa.

Scientific Applications:

• Microbial biodiversity and ecology: Elucidates network structure and co-occurrence modules in environmental microbiomes to study community composition, interactions, and functional dynamics.
• Clinical and human health studies: Analyzes taxon co-occurrences and network modules in clinical metagenomic datasets to investigate associations with clinical conditions and human health.

Methodology:

Utilizes pairwise similarity/distance measures adapted from ecological sample analysis; implements a sparse inverse of a similarity matrix for network construction (extending sparse inverse covariance methods); and evaluates the approach on both real and simulated data.