NetSAM

NetSAM uncovers hierarchical and modular organization in biological networks by performing network seriation and multiscale modularization on weighted or unweighted edge-list representations.

Key Features:

• Edge-list input: Accepts weighted or unweighted edge-list representations as network input.
• Network seriation: Performs network seriation to generate a one-dimensional ordering of nodes.
• Multiscale modularization: Applies multiscale modularization algorithms to identify hierarchical, biologically coherent network modules.
• Correlation-based network construction: Constructs correlation-based networks, such as gene co-expression networks, directly from input matrix data.
• Module–phenotype association testing: Supports statistical testing of associations between identified modules and phenotypes.
• Gene Ontology enrichment: Performs enrichment analysis for Gene Ontology terms linked to identified modules.
• Outputs: Produces ordered gene lists and hierarchical modules for downstream analyses.
• Application domain: Demonstrated on human and mouse protein–protein interaction (PPI) networks to reveal higher-order organizational structure and functionally meaningful modules.

Scientific Applications:

• Hierarchical network organization discovery: Identifies multiscale modular structure and higher-order organization in biological networks.
• Gene co-expression analysis: Builds correlation-based gene co-expression networks from expression matrices for modular analysis.
• Module–phenotype association: Tests associations between network modules and phenotypic variables.
• Functional enrichment profiling: Characterizes modules by Gene Ontology term enrichment.
• Integration of heterogeneous datasets: Provides ordered gene lists and hierarchical modules to align diverse genomic and functional datasets along a network-informed axis.

Methodology:

NetSAM performs network seriation to produce one-dimensional node orderings, applies multiscale modularization algorithms to identify hierarchical modules, constructs correlation-based networks from input matrices (e.g., gene co-expression), and supports module–phenotype association testing and Gene Ontology enrichment analysis.