Increasing certainty in systems biology models using Bayesian multimodel inference.

Area of Science:

• Systems Biology
• Computational Biology
• Mathematical Modeling

Background:

• Intracellular signaling networks are complex, necessitating mathematical models for study.
• Phenomenological approximations are often used, leading to multiple models for the same pathway.
• Model selection and prediction certainty are challenged by multiple, potentially incomplete models.

Purpose of the Study:

• To investigate Bayesian multimodel inference (MMI) for increasing certainty in systems biology predictions.
• To leverage a set of potentially incomplete models for enhanced predictive power.
• To identify mechanisms of subcellular location-specific extracellular-regulated kinase (ERK) activity.

Main Methods:

• Applied Bayesian multimodel inference (MMI) to existing models of the extracellular-regulated kinase (ERK) pathway.
• Evaluated the robustness of MMI predictors to changes in model sets and data uncertainties.
• Utilized MMI to analyze experimentally measured subcellular location-specific ERK activity.

Main Results:

• MMI successfully combined multiple models of the ERK pathway.
• MMI yielded predictors that were robust to model set variations and data uncertainties.
• MMI facilitated the identification of potential mechanisms underlying location-specific ERK activity.

Conclusions:

• Bayesian multimodel inference is a disciplined approach to enhance certainty in systems biology predictions.
• MMI offers a robust method for integrating information from multiple, potentially incomplete pathway models.
• This work demonstrates MMI's utility in dissecting complex signaling dynamics like location-specific ERK activity.