An artificial intelligence system enables robots to perform autonomous scientific experiments, up to 10,000 a day, which could fuel a dramatic leap in the pace of discovery in areas ranging from medicine to agriculture to environmental science.
reported today in microbiology of nature, the team was led by a professor now at the University of Michigan.
That artificial intelligence platform, dubbed BacterAI, mapped the metabolism of two microbes associated with oral health, with no baseline information to begin with. Bacteria consume some combination of the 20 amino acids needed to sustain life, but each species requires specific nutrients to grow. The UM team wanted to know which amino acids the beneficial microbes in our mouths need so they can promote their growth.
“We know almost nothing about most of the bacteria that influence our health. Understanding how bacteria grow is the first step in redesigning our microbiome,” said Paul Jensen, a UM assistant professor of biomedical engineering who was at the University of Illinois when the project began. .
However, figuring out the combination of amino acids that bacteria like is tricky. Those 20 amino acids produce over a million possible combinations, just based on whether each amino acid is present or not. However, BacterAI was able to discover the amino acid requirements for growth of both Streptococcus gordonii and Streptococcus sanguinis.
To find the right formula for each species, BacterAI tested hundreds of amino acid combinations per day, refining their approach and changing the combinations each morning based on the previous day’s results. Within nine days, it was producing accurate predictions 90% of the time.
Unlike conventional approaches that feed labeled data sets into a machine learning model, BacterAI creates its own data set through a series of experiments. By analyzing the results of past trials, you come up with predictions of which new experiments might give you the most information. As a result, he discovered most of the rules for feeding bacteria with less than 4,000 experiments.
“When a child learns to walk, they don’t just watch adults walk and then say ‘Okay, I’ve got it,’ they stand up and start walking. They fumble first and do a little bit of trial and error,” he said Jensen.
“We wanted our AI agent to step up and fall down, have his own ideas and make mistakes. Every day he gets a little better, a little smarter.”
Little to no research has been done on about 90% of bacteria, and the amount of time and resources required to learn even basic scientific information about them using conventional methods is overwhelming. Automated experimentation can dramatically speed up these discoveries. The team performed up to 10,000 experiments in a single day.
But the applications go beyond microbiology. Researchers in any field can pose questions as puzzles for the AI to solve through this kind of trial and error.
“With the recent explosion of mainstream AI in the past few months, many people are unsure what it will bring in the future, both positive and negative,” said Adam Dama, a former Jensen Lab engineer and lead author of the study. . “But to me, it’s very clear that focused AI applications like our project will speed up everyday research.”
The research was funded by the National Institutes of Health with support from NVIDIA.
—————————————————-
Source link