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

Overview

Who: Zachary Cosenza, David E. Block, Keith Baar, Xingyu Chen

Published: June 28, 2023

Where: Engineering in Life Sciences

Key Takeaway: Machine learning can be used to rapidly optimize media

Research Topics: media

Summary

Zachary Cosenza et al use machine learning techniques to design a serum-free growth medium that minimizes cost while maximizing cell growth. Optimizing culture media is challenging because media contains many different ingredients, each with multiple possible concentrations, leading to thousands of possible combinations. The study uses an algorithm that learns to identify the trade-offs between cell growth and medium cost, resulting in a medium with 23% more growth at only 62.5% of the cost of the starting medium. The study uses passage 2 data for the algorithm, which the authors note is a limitation because it does not fully predict the dynamics of cell growth over long passage times. However, using short experimental conditions reduces the time and resources used. In addition, the study indicates that results are cell type dependent, requiring re-optimization for new cell types. Luckily, machine learning can help make re-optimization faster!

Go to Publication

Funding

Cite This Publication

Cosenza, Z., Block, D. E., Baar, K., & Chen, X. (2023). Multi‐objective Bayesian algorithm automatically discovers low‐cost high‐growth serum‐free media for cellular agriculture application. Engineering in Life Sciences, 23(8), e2300005.

Copied!