SUBATOMIC

SUBATOMIC identifies and clusters composite two- and three-node subgraphs in multi-omics interaction networks to produce topologically distinct modules for functional annotation and interpretation of gene regulation and disease-associated pathways.

Key Features:

• Subgraph detection: Identifies composite network subgraphs comprising two to three nodes.
• Subgraph-based clustering: Clusters detected subgraphs into topologically distinct modules.
• Multi-omics integration: Operates on composite networks integrating transcription factor–target, miRNA–target, protein–protein, homologous, and co-functional interactions in Homo sapiens.
• Functional annotation: Assigns Gene Ontology (GO) term annotations to modules.
• Expression integration: Incorporates expression profiles to assess module activity and context-specific responses.
• Module prioritization: Implements two prioritization strategies: GO term enrichment and an activity score reflecting differential expression.
• Context-specific analysis: Applied to study hypoxia responses in three cancer cell lines.
• Inter-module and regulator analysis: Statistically examines connections between modules and their regulators, including miRNAs and transcription factors.
• Module visualization: Produces visual representations of modules to aid interpretation.
• High-throughput output: Generated and functionally annotated 5,586 modules with varied topological, functional, and regulatory characteristics.

Scientific Applications:

• Functional annotation of genes: Derives GO-based functional hypotheses for annotated and uncharacterized genes based on module membership.
• Interpretation of multi-omics networks: Reduces network complexity by producing interpretable modules from dense multi-omics "hairball" networks.
• Condition-specific module discovery: Identifies modules responsive to specific conditions such as hypoxia in cancer cell lines.
• Regulatory architecture analysis: Reveals key regulators (miRNAs and transcription factors) and functionally related modules within a broader network context.
• Module prioritization for follow-up: Uses GO enrichment and activity scores to rank modules for experimental validation or further analysis.
• Analysis of human interaction networks: Applied to composite Homo sapiens networks integrating diverse interaction types.

Methodology:

Detects composite two- and three-node subgraphs and clusters them into topologically distinct modules; functionally annotates and visualizes modules; integrates module data with expression profiles; statistically examines inter-module connections and regulator relationships; applies GO enrichment and computes an activity score reflecting differential expression for module prioritization on a composite Homo sapiens interaction network.