Dynamic optimization of biological networks under parametric uncertainty

Area of Science:

• Systems biology
• Biochemical engineering
• Computational biology

Background:

• Micro-organisms are vital in biochemical, food, and pharmaceutical industries.
• Understanding intracellular biochemical reactions enhances process control.
• Biological networks are key tools in systems biology for modeling cellular processes.
• Dynamic biological systems often involve conflicting objectives, necessitating multi-objective optimization.
• Parametric uncertainty in biological models can lead to constraint violations and inaccurate predictions.

Purpose of the Study:

• To compare uncertainty propagation techniques for dynamic optimization of biological networks.
• To evaluate methods for handling parametric uncertainty in multi-objective optimization problems.
• To assess the robustness of different strategies in biological process optimization.

Main Methods:

• Comparison of three uncertainty propagation techniques: linearization, sigma points, and polynomial chaos expansion.
• Application of these techniques to two case studies: a linear pathway model and a glycolysis-inspired network model.
• Utilizing Monte Carlo simulations to assess constraint violations under uncertainty.

Main Results:

• All tested uncertainty propagation strategies provide robust solutions under parametric uncertainty.
• Sigma points and polynomial chaos expansion demonstrate superior performance in reducing constraint violations.
• Polynomial chaos expansion can directly incorporate prior knowledge of uncertainty distributions.

Conclusions:

• Uncertainty propagation techniques are essential for robust optimization of biological networks.
• Sigma points and polynomial chaos expansion offer a favorable balance between computational efficiency and robustness.
• These methods enhance the reliability of model-based optimization in biological process control.