Multiscale Probabilistic Modeling: A Bayesian Approach to Augment Mechanistic Models of Cell Signaling with...

Area of Science:

• Systems Biology
• Computational Biology
• Biophysics

Background:

• Computational models in systems biology are frequently underdetermined due to limited experimental data.
• Existing biological databases offer observations but often lack direct relevance to specific systems of interest.
• Challenges arise from differing experimental conditions and scales in database measurements.

Purpose of the Study:

• To investigate the utility of generalizing biological databases across differing experimental conditions and scales.
• To enhance the determination and utility of computational models for specific biological systems, particularly cell signaling.
• To quantify the information gained by integrating diverse database measurements into model parameter inference.

Main Methods:

• Development of a novel, multiscale, probabilistic framework for data integration.
• Integration of protein structure data from the Protein Data Bank (PDB) and amino acid sequence data from the Universal Protein Resource (UniProt).
• Application of the framework to parameter inference in dynamic cell signaling models.

Main Results:

• Successful integration of PDB and UniProt measurements significantly improved parameter estimation for cell signaling models.
• The impact of sequence and structure data on model predictions was found to be dependent on parameter sensitivity.
• Demonstrated that protein structure and sequence measurements can effectively inform model parameters.

Conclusions:

• Generalizing biological databases across differing experimental contexts is feasible and beneficial for computational modeling.
• The proposed probabilistic framework offers a robust method for integrating diverse biological data.
• Leveraging readily available protein sequence and structure data can substantially improve the accuracy and utility of systems biology models, especially for dynamic processes like cell signaling.