AI speeds up yeast research for producing oils and pigments

Researchers at Tallinn University of Technology have used artificial intelligence to predict which genes could make oil yeast more efficient at producing lipids and pigments. The new metabolic model could speed up the search for raw materials for food, cosmetics and biofuels, ERR reported.

The international research team, led by Estonian bioengineers, studied Yarrowia lipolytica, a yeast that stores large amounts of oil-like substances in its cells. The scientists wanted to find ways to make it produce valuable pigments called carotenoids as well as lipids.

Lipids naturally account for between one-fifth and half of the yeast’s dry mass. Changing its genes can raise that share to nearly 80 percent and enable the cells to produce carotenoids they do not normally make. But because thousands of metabolic processes are linked, changing even one gene can affect the wider system.

Researchers use genome-scale metabolic models to identify promising gene variants before testing them in the laboratory. Earlier models had gaps in their data on enzyme properties. The team addressed this with an artificial intelligence algorithm called DLKcat, which predicts how almost all Yarrowia lipolytica enzymes behave in different conditions.

Study lead author Juliano Sabedotti de Biaggi, a researcher at Tallinn University of Technology, said the AI accelerated enzyme analysis several hundred times. Calculations that had taken several days for one enzyme could be completed for hundreds of enzymes in less than an hour, he said.

The researchers also collected proteomics data in laboratory experiments and added it to the model. The data show which proteins the yeast produces, and in what quantities, at different stages of its life cycle. This enabled the team to create ecYali5-GEM, a more detailed metabolic model.

The model can help researchers identify when to increase or decrease the activity of a particular gene. A gene that boosts lipid production in an early growth stage may have no effect later in the yeast’s life cycle.