MultiPaths

MultiPaths applies diffusion algorithms to analyze multi-layer biological networks for integrating multi-omics data and interpreting high-throughput biomolecular interaction datasets.

Key Features:

• Two Python packages: Composed of DiffuPy and DiffuPath as independent Python packages for algorithmic application and network-specific integration.
• DiffuPy: Provides a suite of state-of-the-art diffusion algorithms applicable to generic networks for propagating information across nodes.
• DiffuPath: Extends diffusion algorithms specifically to multi-layer biological networks to integrate multiple data modalities.
• Multi-layer network integration: Supports construction and analysis of multi-layer networks that combine different biomolecular interaction modalities.
• Pathway database–derived networks: Generates networks from various pathway databases and enables their harmonization across disparate resources.
• Multi-omics analysis: Applied diffusion experiments on multi-omics datasets to uncover patterns and relationships across modalities.
• Predictive performance enhancement: Harmonized networks from disparate databases improve predictive performance compared to individual resources.

Scientific Applications:

• High-throughput data interpretation: Interprets and contextualizes results from high-throughput screening using diffusion on biomolecular interaction networks.
• Multi-omics integration: Integrates and relates diverse omics modalities within a unified multi-layer network framework.
• Network-based prediction and prioritization: Enhances prediction and prioritization of entities by propagating signals through harmonized pathway-derived networks.
• Comparative evaluation of network resources: Assesses the impact of harmonized versus individual pathway databases on predictive performance.

Methodology:

Application of diffusion algorithms (information propagation) via DiffuPy and DiffuPath to networks generated from pathway databases, harmonization of disparate databases, and diffusion experiments on three independent multi-omics datasets.