MAGNET

MAGNET generates and analyzes gene-gene coexpression and weighted protein-protein interaction (PPI) networks by integrating user-supplied experimental data with public -omics and high-throughput proteomic datasets to assess molecular interaction plausibility.

Key Features:

• Coexpression network construction: Represents genes as nodes and expression correlations as edges, filtering networks to highlight significant gene-gene relationships from user-supplied expression data.
• Weighted PPI network scoring: Weights PPI edges using a logistic regression model that incorporates tissue-specific gene expression levels, sub-cellular localization information, co-clustering tendencies of interacting proteins, and frequency of observed interactions.
• Edge weighting and quantitative assessment: Provides quantitative scores for interaction plausibility to support inference of functional associations and regulatory mechanisms.
• Multi-omics integration: Integrates user experimental measurements with public -omics and high-throughput proteomic datasets to inform network construction and prediction.
• Network filtering: Applies significance-based filtering to prioritize biologically relevant gene-gene and protein-protein relationships.

Scientific Applications:

• Gene expression pattern analysis: Analyzes coexpression patterns across different conditions or tissues using user-provided expression data.
• Interaction inference: Quantitatively assesses PPI plausibility and infers potential functional associations and regulatory mechanisms from weighted networks.
• Disease and target discovery: Identifies novel interactions and pathways relevant to disease modeling and therapeutic target discovery by combining experimental and public -omics data.

Methodology:

Constructs coexpression networks with genes as nodes and correlation-based edges, applies network filtering for significance, and builds weighted PPI networks using a logistic regression model that incorporates tissue-specific expression, subcellular localization, co-clustering, interaction observation frequency, and integration of public -omics and high-throughput proteomic datasets with user data.