A Computational Framework for Optimal and Model Predictive Control of Stochastic Gene Regulatory Networks

Area of Science:

• Systems Biology
• Synthetic Biology
• Control Theory

Background:

• Designing controllers for cellular functions faces challenges in managing molecular noise.
• Accurate and efficient solution of the Chemical Master Equation is a bottleneck for model-based control of stochastic biomolecular systems.

Purpose of the Study:

• Develop a framework for optimal and Model Predictive Control of stochastic gene regulatory networks.
• Address limitations in computational efficiency and precise control over cell population behavior.
• Provide robust handling of intrinsic molecular noise in biological systems.

Main Methods:

• Utilized an efficient approximation of the Chemical Master Equation via Partial Integro-Differential Equations.
• Implemented an adjoint-based optimization method for enhanced control.
• Applied the framework to stochastic gene regulatory networks.

Main Results:

• Achieved high computational efficiency in control system design.
• Demonstrated precise control over the probability density function for complex cell population behaviors, including bimodality.
• Showcased robust handling of intrinsic molecular noise.

Conclusions:

• The developed framework offers significant advantages for Cybergenetics and Synthetic Biology.
• Enables fine-tuning of cell populations for emergent properties and dynamic tracking.
• Provides an effective approach for model-based control of noisy biomolecular systems.