CAZAI: CAZyme Specificity Prediction Using AI

Bernard Henrissat – Professor, Department of Biotechnology and Biomedicine, DTU
Grant amount: DKK 14,995,489

Carbohydrate processing enzymes (CAZymes) play important roles in our bodies and nature but also hold great potential within sustainable green production of energy and materials, as an alternative to fossil-fuel-based production. This project will tackle the challenge of predicting the substrate specificity of carbohydrate processing enzymes by developing new deep learning methods in a loop between experiments (determining the substrate specificity of enzymes) and modelling (training of deep learning methods on this data). The committee acknowledges that this is a major unsolved scientific challenge (even enzymes with highly similar 3D structures may have very different substrate specificities) but believe that this consortium of researchers is well-poised to address it.

The project will contribute to fundamental methods development in an area which is broadly applicable across both health and sustainability (predicting and understanding protein function). At the same time, a method which can predict specificity directly from sequence would be highly impactful for discovery and optimization of enzymes, which remains to be a stronghold of the Danish biotech sector. Ultimately, new improved carbohydrate converting enzymes could accelerate the green transition. 

Project participants
Bernard Henrissat
Professor, Department of Biotechnology and Biomedicine, DTU

Ole Winther
Professor, Department of Biology (KU) & Department of Applied Mathematics and Computer Science (DTU-Compute), KU & DTU

Renaud Vincentelli
Research Scientist, Architecture et Fonction des Macromolecules Biologiques, Aix-Marseille University, France