Communities formed by human intestinal microbes can now be predicted more accurately with a new computer model developed in a collaboration between biologists and engineers, led by the University of Michigan and the University. of Wisconsin.
The making of the model also suggests a route to ascent from the 25 microbe species explored to the thousands that may be present in human digestive systems.
“Every time we increase the number of species, we get an exponential increase in the number of possible communities,” said Alfred Hero, the John H. Holland Distinguished University Professor of Electrical Engineering and Computer Science at University of Michigan and co -author of the study journal eLife.
“That’s why it’s so important that we can extrapolate from data collected in several hundred communities to predict the behaviors of millions of communities that we don’t see.”
While research continues to reveal the many ways that microbial communities influence human health, probiotics have often not followed the hype. We don’t have a good way of predicting how the introduction of a strain will affect the existing community. But machine learning, an artificial intelligence technique in which algorithms learn to make predictions based on data sets, can help change that.
The problems of this dimension require a complete overhaul of how we model community behavior. ”
Mayank Baranwal, adjunct professor of systems and control engineering at the Indian Institute of Technology, Bombay, and co-first author of the study
He explained that the new algorithm could map the entire landscape of 33 million possible communities in minutes, compared to the days to months required for conventional ecological models.
Microbial zinc cities
Key to this big step was Ophelia Venturelli, assistant professor of biochemistry at the University of Wisconsin and co -author of the study. Venturelli’s lab runs experiments on microbial communities, storing them in low-oxygen environments that mimic the surroundings of the mammalian gut.
His team has created hundreds of different communities with germs that are widespread in the large human intestine, mimicking the healthy state of the gut microbiome. They then measured how these communities have evolved over time and the concentrations of important health-related metabolites, or chemicals produced as microbes break down foods.
“Metabolites are made in very high concentrations in the intestines,” Venturelli said. “Some are beneficial to the host, like butyrate. Some have more complex interactions with the host and gut community.
The machine learning model enabled the team to design communities with preferred metabolite profiles. This type of control can help doctors find ways to treat or protect against diseases by identifying the right germs.
Feedback for faster model construction
While research into a person’s microbiome has a long way to go before it can provide this type of intervention, the approach the team has developed can help get there faster. Machine learning algorithms are usually performed using a two-step process: training data collection, and then algorithm training. But the feedback step added by Hero and Venturelli’s team provides a template for rapid improvement on future models.
The Hero team initially trained the machine learning algorithm on an existing set of data from Venturelli’s lab. The team then used the algorithm to predict the evolution and metabolite profiles of the new communities established and tested by Venturelli’s team in the lab. While the model was very good overall, some of the predictions identified weaknesses in the model’s performance, with Venturelli’s group highlighting the second phase of the experiments, closing the feedback loop.
“This new modeling approach, combined with the speed with which we can test new communities in Venturelli’s lab, could enable the design of beneficial microbial communities,” said Ryan Clark, co- first author of the study, who was a postdoctoral researcher in Venturelli’s lab when he ran the microbial experiments. “It’s much easier to optimize for making multiple metabolites at once.”
The group resides in a long short-term memory neural network for machine learning algorithms, which are good for sequence prediction problems. However, like most machine learning models, the model itself is a “black box.” To determine what factors were present in its predictions, the team used the mathematical map generated by the trained algorithm. It reveals how each type of microbe affects the abundance of the others and what types of metabolites it supports. They can use these relationships to design communities that should be studied through modeling and follow-up experiments.
The model can also be used in a variety of microbial communities beyond medicine, including accelerating the breakdown of plastics and other materials for cleaning the environment, making valuable compounds for bioenergy applications, or remediation. in plant growth.
This study was supported by the Army Research Office and the National Institutes of Health.
Hero is also R. Jamison and Betty Williams Professor of Engineering, and a professor of biomedical engineering and statistics. Venturelli is also a professor of bacteriology and chemical and biological engineering. Clark is now a senior scientist at Nimble Therapeutics. Baranwal is also a scientist in the data and decision sciences division at Tata Consultancy Services Research and Innovation.
Baranwal, M., and so on. (2022) Repetitive neural networks enable the design of multifunctional synthetic human gut microbiome dynamics. eLife. doi.org/10.7554/eLife.73870.