Green transition

Researchers will map the dynamics of enzymes with Denmark's first AI supercomputer

The Novo Nordisk Foundation and the Danish Centre for AI Innovation have selected the first research projects to gain access to the AI supercomputer ‘Gefion’. Among them is a DTU research project that will map the dynamics of enzymes and develop enzymes that can convert CO2 into sustainable fuels and chemicals.

Photo: Novo Nordisk Foundation.
DTU researchers are among the first to gain access to Denmark's first AI supercomputer, 'Gefion.' The researchers will use Gefion's extraordinary computational power to map the dynamics of enzymes. Photo: Novo Nordisk Foundation.

Wednesday 23 October 2024

Katrine Damkjær

A high-profile research project from DTU is among the first six projects that have made it through the eye of the needle and will have access to the new AI supercomputer funded by the Novo Nordisk Foundation and the Danish Export and Investment Fund established in collaboration with the US tech company NVIDIA.
 
In the project, researchers from different departments and centers at DTU will use a new type of machine learning to perform large-scale simulations of enzymes to map how enzymes move. By understanding their dynamics, the researchers can develop more efficient enzymes that can convert captured CO2 into sustainable fuels and chemicals.
 
"Access to Gefion will allow us to study a class of very interesting enzymes that can convert CO2 into sustainable fuels and chemicals. This can be done with unprecedented accuracy and speed, allowing us to identify new, efficient enzymes for converting CO2 into sustainable fuels such as methanol," says DTU Professor Tejs Vegge and Director of the Pioneer Center CAPeX.

Together with Senior Researcher Carlos G. Acevedo-Rocha and a broad team of DTU researchers, Tejs Vegge is behind the project. Using different interdisciplinary approaches, they are working on cracking the code to deliver green fuels at a large scale.

In the future, using enzymes to convert captured CO2 could be a sustainable and efficient supplement to the man-made power-to-X processes we use today.
 

Green vision

The fact that DTU is one of the first universities to gain access to the supercomputer emphasizes the university's vision of making a positive difference to society and the green transition, states Christine Nellemann, Dean of Sustainability, Diversity, Inclusion and International Collaborations at DTU.

Gefion's extraordinary computing power opens up completely new opportunities in the research of green technologies that can help accelerate our transition to renewable energy.
 
"That our researchers have the opportunity to run large datasets and simulations from excellent research initiatives from DTU at Gefion not only means that we can learn more about nature's own processes for converting CO2, but also that we can deliver results faster, which is crucial if we are to deliver solutions to counteract global warming."
Christine Nellemann Dean of Sustainability, Diversity, Inclusion and International Collaborations at DTU.
The enzyme formate dehydrogenase, shown as spirals in rainbow colors. The enzyme converts CO2 into the product formate, represented as small white balls. The cofactor NAD+, which enables the reaction, is shown with gray balls. The blue represents many small water molecules surrounding and dissolving the enzyme. The figure is created using PDB code 2GO1 by Filippova et al. 2005 and The PyMOL Molecular Graphics System, Version 3.0 Schrödinger, LLC.

Computing power will accelerate basic research

Enzymes already play a major role in our everyday lives as small machines that help speed up chemical reactions. Not only in our bodies, but also in the production of food, medicine, sustainable biomaterials and chemicals.
 
Enzymes also play a central role in many biological processes involving CO2 conversion in nature. By understanding these processes, we can develop efficient enzymes that can convert captured CO2 into sustainable fuels such as methanol.
 
With Gefion, the interdisciplinary team can simulate a much larger number of atoms and take into account how enzymes are affected by fluids, which is crucial to accurately predict their function and movements over time.
 
"Designing high-performance enzymes is complex and very difficult. It requires extraordinary computing power, and it requires us to test the calculated enzyme designs in the lab and learn from the results. Performing experiments and tests in the lab is very important to validate the supercomputer simulations and calculated predictions. This is crucial when using artificial intelligence to accelerate research," says DTU Senior Researcher Carlos G. Acevedo-Rocha.  
 

Facts

About the research project

The DTU research project AIM4CO2: ‘Unravelling CO2 reduction in Non-Metal Formate Dehydrogenase (FDH) using Machine-Learned Force Fields’ will be one of the first research projects to run tests on the Gefion AI supercomputer.

With the project, DTU researchers will perform unprecedented large-scale simulations that are only possible on a supercomputer like Gefion to identify new variants of FDH that can be used to increase the rate of CO2 reduction to formate and perhaps even to methanol, e.g. for sustainable shipping fuels.

The research is an interdisciplinary collaboration between DTU Energy and DTU Biosustain and is carried out as part of CAPeX - Pioneer Centre for Accelerating P2X Materials Development.

The research project is led by Professor Tejs Vegge from DTU Energy and Senior Researcher Carlos G. Acevedo-Rocha from DTU Biosustain, Novo Nordisk Foundation Center for Biosustainability. Additionally, the research team consists of Postdoc Bardi Benediktsson, Associate Professor and PhD Arghya Bhowmik, Scientific Software Developer and PhD Steen Lysgaard, Scientific Software Developer and PhD Jonas Busk, as well as DDSA PhD Fellow Niklas Gesmar Madsen.

The research project has been given a total of 3 months to use Gefion.

Contact

Tejs Vegge Professor, Head of Section Phone: +45 45258201 Mobile: +45 51641787

Carlos G. Acevedo-Rocha Senior Researcher