Accelerating cell culture media development using Bayesian optimization-based iterative experimental design.

Area of Science:

• Biotechnology
• Life Sciences Research
• Computational Biology

Background:

• Optimizing complex biological systems, such as cell cultures, is time-consuming and resource-intensive.
• Traditional methods like Design of Experiments (DOE) can require numerous trials for media development.
• Developing specialized media is crucial for maintaining cell health and enhancing bioproduction.

Purpose of the Study:

• To accelerate cell culture media development using a Bayesian Optimization (BO)-based iterative framework.
• To demonstrate the framework's efficacy in improving cell viability and recombinant protein production.
• To showcase the framework's adaptability and efficiency compared to conventional methods.

Main Methods:

• Application of a Bayesian Optimization iterative framework for experimental design.
• Optimization of media compositions for human peripheral blood mononuclear cell culture.
• Optimization of recombinant protein production in K.phaffii cultivations.
• Utilizing an exploration-exploitation trade-off within the iterative process.

Main Results:

• Identification of novel media compositions that enhance human peripheral blood mononuclear cell viability and distribution.
• Significant improvements in recombinant protein production yields in K.phaffii.
• Achieved optimized conditions using 3-30 times fewer experiments than estimated for standard DOE.
• Demonstrated extensibility through transfer learning for incorporating additional design factors.

Conclusions:

• The BO-based iterative framework substantially accelerates media development and optimization in biotechnology.
• This approach offers a more efficient and resource-sparing alternative to traditional experimental design methods.
• The framework's flexibility and demonstrated success highlight its potential for complex biological system optimization.