Abstract
Microbial media are composed of various nutrients including specific chemicals and natural raw materials and their extracts. Since there are many different types of microbial culture media, optimization is difficult with only experimental testing. Here, two different approaches will be introduced: metabolomics profiling, and optimization assisted with machine learning. In metabolomics profiling, data from non-targeted comprehensive analyses can be used to find significant components in raw materials and their extracts with statistical calculat ion and machine learning. Optimization with assisted machine learning consists of data sampling, validating machine learning, and maximum value search by optimization algorithms. This optimization method optimized a medium with 31 components for protein expression by Escherichia coli.
