Photo: Mikal Schlosser

Enzyme research using graphics cards

Monday 06 Oct 14

Supported by Lundbeck

The project—‘In Silico, Prediction of the Substrate Specificity of Sirtuins in Combination with Chemical Synthesis and Biochemical Profiling’—is supported by the Lundbeck Foundation which has invested DKK 1.25 million.

The grant will be used to fund a PhD research fellowship. The New Carlsberg Foundation has given DKK 100,000 towards the purchase of graphics cards. DTU Chemistry is co-financing the project.

Researchers from DTU Chemistry are trying to penetrate the human body’s ‘control centre’ by studying a group of enzymes using computer modelling and experiments.

Two researchers are to investigate enzymes that could potentially be used in cancer treatment. Normally, this kind of project begins in the lab, but on this occasion, the two researchers have decided to allow a computer with four graphics cards—every gamer’s dream scenario—to generate ideas on how the enzymes are constructed. Only after the modelling phase has been completed will they move to the lab to test the computer-generated hypotheses. The project, which may lead to the manufacture of new drugs, is supported by the Lundbeck Foundation.

A class of enzymes called sirtuins (SIRT 1-7) are proving particularly interesting to scientists, as they are believed to play a role in metabolism and the regulation of our genes. They may therefore have an important part to play in the fight against cancer, among other things. Some of the enzymes help to ensure that the right genes in the DNA strand are decoded at the right moment, but their exact structure and function are as yet unknown. The PhD project anchored at DTU Chemistry seeks to address this problem.

The project is devised by Associate Professor Peter Fristrup, DTU Chemistry—and Professor Christian Adam Olsen, who this spring moved from DTU Chemistry to the Center for Biopharmaceuticals, University of Copenhagen. The two researchers have previously used computer modelling to explain experimental observations.

This time, they have turned the method on its head, beginning with computer-generated ideas and hypotheses in order to validate them by means of chemical synthesis and enzymology.

“We want to examine processes taking place deep within the cell ‘control centre’ to determine which molecules are converted by which enzymes. The basic knowledge we obtain will hopefully be used actively in the design and manufacture of new drugs,” says Christian Adam Olsen, who works in the fields of organic chemistry and chemical biology.

Graphics cards for research
A large part of the computing power must come from graphics cards originally developed for computer games.

“Graphics cards have become incredibly fast and inexpensive precisely because they are used in the popular gaming industry and we expect them to prove highly suitable for the specific type of modelling we have in mind,” says the team’s modelling expert, Peter Fristrup.

At the moment he is testing a computer with four graphics cards that will consume 800 watts. Although this sounds like a lot, it is only about one tenth of what a cluster of traditional computers with equivalent performance would use.

He and Christian have also found the perfect PhD student for the job.

“We needed a bit of a multi-artist—someone capable of doing the complicated computer calculations and with a knowledge of organic synthesis and biochemical reactions who could work in the lab —an unlikely combination, but we found the perfect candidate in Rita Colaço from Portugal, who will begin working on the project on 1 November,” says Peter Fristrup.

Article in DTUavisen no. 8, October 2014.