Integrated In Vitro/In Silico Uncertainty Quantification Method for Protein Crystallization Models

Area of Science:

• Biotechnology
• Chemical Engineering
• Crystallization Science

Background:

• Protein crystallization is complex, hindering biomanufacturing process intensification.
• Current computational models require accurate parameter estimation for experimental validation.
• Nonlinear model structures and inaccurate process analytical technology impede effective parameter estimation.

Purpose of the Study:

• To develop and validate a model-driven parametrization methodology for antisolvent batch protein crystallization.
• To improve the accuracy of computational models for bioprocess development.
• To address challenges in parameter estimation for nonlinear crystallization systems.

Main Methods:

• Developed an experimentally validated, model-driven parametrization methodology for batch protein crystallization.
• Employed global sensitivity analysis to identify parameter identifiability and optimal measurement points.
• Utilized Approximate Bayesian Computation and Monte Carlo simulations for parameter uncertainty estimation and propagation.

Main Results:

• Successfully estimated parameters for a batch protein crystallization system with limited offline data.
• Validated the methodology under new experimental conditions, demonstrating robustness.
• Quantified parametric and output uncertainties, highlighting the need for nonlinear model-specific methods.

Conclusions:

• The presented methodology enhances the reliability of computational models in protein crystallization.
• Accurate parameter estimation is crucial for advancing biomanufacturing through process intensification.
• Tailored methodologies are essential for handling parameter uncertainty in nonlinear systems.