Systematic data-driven genome-scale metabolic model reduction for bioprocess modeling: CHO culture case study

Area of Science:

• Systems Biology
• Metabolic Engineering
• Biotechnology

Background:

• Genome-scale metabolic models (GEMs) offer mechanistic insights into cellular metabolism but are often too large and underdetermined for dynamic simulations.
• Existing model reduction methods struggle with experimental uncertainty, simplified rate estimations, or manual assumptions, limiting their robustness and scalability.

Purpose of the Study:

• To develop a novel, uncertainty-aware pipeline for reducing GEMs using metabolomics data.
• To generate compact GEMs suitable for dynamic bioprocess simulation and digital twin integration.

Main Methods:

• A metabolomics-driven reduction pipeline integrating Bayesian flux estimation to incorporate uncertainty from exo-metabolomics data.
• Application to time-course exo-metabolomics data from 12 fed-batch Chinese Hamster Ovary (CHO) cell cultures.

Main Results:

• A single reduced GEM was generated that remained feasible across all tested conditions.
• The reduced model accurately reproduced observed extracellular fluxes.
• The model preserved broad metabolic functionality despite relying solely on extracellular data.

Conclusions:

• The developed pipeline provides a systematic, uncertainty-aware framework for generating compact GEMs.
• This approach enhances the utility of GEMs for dynamic bioprocess simulation and digital twin applications.
• The method demonstrates the predictive power of extracellular metabolomics data for metabolic model reduction.