Science
Newswise – Reactive oxygen species (ROS) are unstable molecules produced by organisms during their natural function. These functions include respiration and the conversion of food into energy. During industrial processes, yeast is often exposed to high levels of ROS. This exposure can limit the amount of high-value fatty acids or other biological organisms that yeast can make. Some yeasts are naturally resistant to oxidative stress. However, scientists have limited knowledge of which species are resistant and how to protect themselves. Here, scientists used machine learning to analyze genetic blueprints of hundreds of yeasts. This analysis allowed us to identify which gene groups are most important for ROS resistance.
Impact
Identifying genetic factors associated with oxidative stress resistance can help scientists understand how these functions evolved. The study highlights genes that scientists can target and create more robust yeast strains for industrial use. These applications include the creation of biofuels or biological organisms. Furthermore, researchers may be able to use this information to combat yeast infections. This work also provides a framework for using machine learning to identify genes associated with the characteristics of many species.
summary
Scientists at the Great Lakes Bioenergy Research Center tested 285 yeasts across the Saccharomycotina sublineage to assess the variance of the group of ROS resistance. We then trained machine learning models to identify the genes that were most important for resistance. This model identified two groups of important genes. One of these groups helps in building and maintaining cell walls. Other groups consist of reductase genes that produce enzymes that neutralize ROS.
To test the predictions, scientists conducted two experiments. First, they took a species that was very sensitive to ROS, K. lactis. The researchers gave an extra copy of the reductase gene to increase the production of the associated enzyme. Modified yeast was resistant to ROS. The researchers then deleted two genes that caused cell wall structure from the second species. Saccharomyces cerevisiae. This deletion made the yeast more susceptible to ROS stress. These experiments show that a link can be drawn between machine learning models prediction and benchtop biology.
Funds
The project was supported by the National Science and Science Environmental Research Program, the U.S. Department of Energy, and the Center for Bioenergy Research in the Great Lakes. National Science Foundation. American Institute of Agriculture and Agriculture. National Institutes of Health/Industrial Diseases Institute. and the Burrows Welcome Fund.
Journal Links: Nature Communications, 16, 1–15. (2025)
