Multi-objective Bayesian algorithm automatically discovers low-cost high-growth serum-free media for cellular...

Area of Science:

• Biotechnology
• Computational Biology
• Cell Culture Optimization

Background:

• Optimizing cell culture media is crucial for biopharmaceutical production.
• Balancing cost and cell growth presents a significant challenge in media development.

Purpose of the Study:

• To apply multi-information source Bayesian optimization for simultaneous cost minimization and cell growth maximization.
• To identify optimal serum-free media formulations for C2C12 myoblast growth.

Main Methods:

• Utilized a multi-objective Bayesian optimization technique with a hyper-volume improvement acquisition function.
• Employed sequential experimental batches with custom media, guided by Bayesian criteria.
• Incorporated multiple assays to capture diverse cellular growth dynamics.

Main Results:

• Identified media formulations yielding significantly higher C2C12 cell growth compared to controls.
• Discovered a medium with 23% increased growth at 62.5% of the control cost.
• Observed sustained cell growth beyond the study period, validating the model's predictive accuracy.

Conclusions:

• The Bayesian optimization approach effectively navigates the trade-off between media cost and cell growth.
• This method enables the discovery of high-performance, cost-effective cell culture media.
• The modeling approach accurately predicts cell growth even with limited experimental data.