AI meets enzyme research

An interdisciplinary team led by Prof. Holger Gohlke, a member of JARA-CSD, has developed a novel AI model capable of predicting enzyme functions based on their structure. The research project, carried out in close collaboration with several partner institutions, was conducted using the JUWELS supercomputer at the Jülich Supercomputing Centre.

Proteins and enzymes form the basis of all life. Understanding their functions is crucial for biotechnological and medical innovation. While the prediction of their three-dimensional structures has advanced significantly thanks to AI tools such as AlphaFold, accurately determining their molecular functions has remained a challenge.

This is where the new model, TopEC, comes into play: it analyses over 250,000 protein structures from existing databases and identifies enzyme functions with high precision. This was made possible by an innovative approach that focuses solely on the atoms closest to an enzyme’s active site. As a result, the training speed on the JUWELS supercomputer increased considerably, while prediction accuracy was improved.

Further information is available on the Forschungszentrum Jülich website: https://www.fz-juelich.de/en/ias/jsc/news/news-items/news-flashes/2025/juwels-in-action-researchers-develop-model-to-predict-enzyme-function-using-supercomputer