Integrating literature-constrained and data-driven inference of signalling networks

Area of Science:

• Systems Biology
• Computational Biology
• Bioinformatics

Background:

• Experimental methods generate large signal transduction datasets.
• Mathematical modeling of these datasets faces challenges with purely data-driven or literature-constrained approaches.
• Existing methods struggle with scalability, interpretability, and incomplete network information.

Purpose of the Study:

• To present an efficient approach for inferring signaling networks from perturbation experiments.
• To integrate literature-constrained and data-driven methods for improved network inference.
• To provide an R package, CNORfeeder, for practical application of the method.

Main Methods:

• The CNORfeeder approach extends existing networks with data-derived links.
• It employs various inference methods to integrate new links.
• Protein physical interaction data is used to guide and validate the integration process.

Main Results:

• CNORfeeder was applied to a growth and inflammatory signaling network.
• The method achieved a superior data fit in HepG2 cells (human liver cancer).
• Potential missing pathways in the signaling network were identified.

Conclusions:

• CNORfeeder offers an efficient way to infer signaling networks by combining literature and experimental data.
• The R package facilitates the integration of diverse data types for robust network reconstruction.
• The approach enhances biological understanding by improving model accuracy and revealing novel pathway components.