Researchers unveil their CATNIP online platform to promote greater use of biocatalysts; fosters greener chemistry

Scientists at the University of Michigan and Carnegie Mellon University have unveiled CATNIP, an open-access platform that uses machine learning to match enzymes with chemical substrates. The system removes a key barrier to biocatalysis and promises to accelerate the use of more sustainable methods in synthetic chemistry

A team of researchers from the University of Michigan at Ann Arbor and Carnegie Mellon University. Pittsburgh, Pennsylvania, has reported the creation of a novel digital platform designed to make greener chemistry more accessible. The development was supported by the United States National Science Foundation with the open access tool removing a principal barrier to the wider adoption of biocatalysis by helping scientists to identify more effectively which enzymes can transform specific chemical compounds.

Biocatalysts – commonly known as enzymes – are proteins that have evolved to perform intricate chemistry with remarkable efficiency. They typically catalyse reactions in water and at room temperature, which eliminates the need for toxic solvents, costly reagents or energy inputs making them attractive alternatives to traditional, harsher chemical processes. However, their natural selectivity has been both a strength and a limitation, since enzymes usually act only upon the particular substrates they encounter in biological systems. For chemists, this narrow specificity has posed challenges when trying to extend the use of biocatalysis to a broader range of molecules.

“Biocatalysis offers a more sustainable way to build molecules. It can also give us access to molecules that we could not build using traditional chemical methods,” said Dr. Alison Narayan, professor of chemistry at the University of Michigan College of Literature, Sciences and the Arts and research professor at the Life Sciences Institute.

“But most of the known substrates for these biocatalysts come from nature which is just a very small subset of the molecules that chemists work with,” she added.

Narayan and her colleagues set out to bridge this gap by systematically exploring the relationship between proteins and potential substrates. The project began with an ambitious attempt to test the compatibility of enzymes and molecules on a large scale. Focusing on one enzyme family, Dr. Alexandra Paton, a postdoctoral researcher in Narayan’s laboratory, designed a high-throughput reaction platform that enabled the team to screen more than 100 different substrates against each protein across the entire family.

“We discovered hundreds of novel connections between chemical space and protein space and built this diverse dataset.

“That is when we began to think more broadly about what we could build with all this data,” said Paton, who is now assistant professor of chemistry at the University of Rochester, New York.

The dataset became the foundation for collaboration with Dr. Gabe Gomes, assistant professor of chemical engineering and chemistry at Carnegie Mellon University. By applying machine learning approaches, Gomes’ laboratory constructed a predictive model that could link enzymes to compatible chemical transformations.

The result was CATNIP, an open-access online platform described by the authors as an enzyme recommender system. Chemists can input a starting compound and receive a ranked list of enzymes from the studied protein family that are most likely to catalyse the desired transformation.

Alternatively, the platform can be queried in reverse to determine which substrates a given enzyme might process. Gomes’ team compared this predictive ability to the functionality of a web search engine, in which the most promising answers are placed at the top of the results.

“It is a great starting model to enable synthetic campaigns using biocatalysts,” Paton said. She added that efforts are already underway to expand the database to include additional enzyme families which would further extend the platform’s utility for the wider chemical community.

The research received additional support from the Novartis Global Scholars Program, the Camille Dreyfus Teacher Scholar Award and the University of Michigan. Other contributors included Jonathan Perkins and Nicholas Cemalovic from the University of Michigan and Thiago Reschützegger from the Federal University of Santa Maria, Brazil.

For further reading please visit: 10.1038/s41586-025-09519-5