Integrating metabolome dynamics and process data to guide cell line selection in biopharmaceutical process...

Area of Science:

• Biotechnology
• Bioprocessing
• Metabolomics

Background:

• Mammalian cell culture for therapeutic antibody production is complex and requires careful cell line selection.
• Selecting stable, high-performing cell lines is crucial for efficient biomanufacturing.
• Integrating biological data with process data can optimize cell line development.

Purpose of the Study:

• To accelerate the selection of high-performing cell lines for monoclonal antibody production.
• To apply machine learning for integrating dynamic metabolomics and process data.
• To predict cell line performance and product titer early in the cultivation process.

Main Methods:

• Utilized micro-bioreactor scale (Ambr®15) for industrial monoclonal antibody development.
• Applied a machine learning approach to integrate time-varying process and metabolomics data.
• Exploited the temporal dynamics of metabolic phenotypes for predictive modeling.

Main Results:

• Successfully predicted cell line performance from early process timepoints.
• Accurately estimated product titer at late process timepoints.
• Identified relationships between metabolic mechanisms and product titer, enabling biomarker discovery.

Conclusions:

• Data-driven metabolic understanding facilitates early identification of high-performing cell lines.
• The machine learning approach provides valuable insights into cell physiological changes.
• Identified key metabolites as potential biomarkers for commercially relevant phenotypes.