Machine learning is transforming all areas of life sciences and industry, but it’s typically limited to a few users and scenarios. A team of researchers from the Max Planck Institute for Terrestrial Microbiology led by Tobias Erb has developed METIS, a modular software system for the optimization of biological systems. The research team demonstrates its ease of use and versatility with a variety of biological examples.
Although biological systems engineering is truly indispensable in biotechnology and synthetic biology, today machine learning has become useful in all areas of biology. However, it is obvious that the application and improvement of algorithms, calculation procedures consisting of lists of instructions, are not easily accessible. Not only are they limited by programming skills, but often also by insufficient experimentally labeled data. At the intersection of computational and experimental work, there is a need for effective approaches to bridge the gap between machine learning algorithms and their applications for biological systems.
Today, a team from the Max Planck Institute for Terrestrial Microbiology led by Tobias Erb has succeeded in democratizing machine learning. In their recent publication in “Nature Communications”, the team presented with collaboration partners from the INRAe Institute in Paris, their METIS tool. The application is built in such a versatile and modular architecture that it does not require computer skills and can be applied on different biological systems and with different laboratory equipment. METIS stands for Machine Learning Guided Experimental Trials for System Improvement and is also named after the ancient goddess of wisdom and craftsmanship Μῆτις, lit. “wise counsel”.
Less data required
Active learning, also known as optimal experimental design, uses machine learning algorithms to interactively suggest the next set of experiments after being trained on previous results, a valuable approach for laboratory scientists wet, especially when working with a limited amount of experimentally labeled data. But one of the main bottlenecks is the experimentally generated labeled data in the lab, which is not always high enough to train machine learning models.
While active learning already reduces the need for experimental data, we went a step further and looked at various machine learning algorithms. Encouragingly, we found an even less data-dependent model.”
Amir Pandi, one of the main authors of the study
To show the versatility of METIS, the team used it for a variety of applications, including optimization of protein production, genetic constructs, combinatorial engineering of enzyme activity, and complex CO2 fixation metabolic cycle named CETCH. For the CETCH cycle, they explored a combinatorial space of 1025 conditions with only 1000 experimental conditions and reported the most efficient CO2 fixation cascade described to date.
Optimize biological systems
In application, the study provides new tools to democratize and advance current efforts in biotechnology, synthetic biology, genetic circuit design, and metabolic engineering. “METIS allows researchers to optimize their already discovered or synthesized biological systems,” says Christoph Diehl, co-lead author of the study. “But it’s also a combinatorial guide to understanding complex interactions and hypothesis-based optimization. And what’s probably the most exciting benefit: it can be a very useful system for prototyping new-in-nature systems. .”
METIS is a modular tool that works like Google Colab Python notebooks and can be used via a personal copy of the notebook on a web browser, with no installation, registration or need for local computing power. The materials provided in this work can guide users to customize METIS for their applications.
Pandi, A. et al. (2022) A versatile active learning workflow for optimizing genetic and metabolic networks. Communication Nature. doi.org/10.1038/s41467-022-31245-z.