MultiPaths: a Python framework for analyzing multi-layer biological networks using diffusion algorithms

Area of Science:

• Computational Biology
• Bioinformatics
• Systems Biology

Background:

• High-throughput screening generates large, complex biological datasets that are challenging to interpret.
• Knowledge-driven approaches using biological networks offer a solution for analyzing this data.
• Manual investigation of large biological networks is impractical, necessitating computational methods.

Purpose of the Study:

• To present MultiPaths, a novel Python framework for network analysis.
• To provide tools for applying diffusion algorithms to generic and multi-layer biological networks.
• To demonstrate the framework's utility in integrating multi-omics data and improving predictive performance.

Main Methods:

• Development of two Python packages: DiffuPy for generic network diffusion algorithms and DiffuPath for multi-layer biological networks.
• Implementation of a command-line interface, reproducible examples, and documentation for user-friendliness.
• Application of diffusion experiments on multi-omics datasets using networks from pathway databases.

Main Results:

• Multi-layer networks effectively integrate multiple biological data modalities.
• Harmonized networks generated from disparate databases enhance predictive performance compared to individual resources.
• The MultiPaths framework facilitates the analysis of complex biological data.

Conclusions:

• The MultiPaths framework, including DiffuPy and DiffuPath, offers a robust solution for interpreting high-throughput biological data.
• Multi-layer network analysis is crucial for integrating diverse biological information.
• Harmonizing data sources through network construction improves the predictive power of computational analyses.