Two Prestigious Grants from European Research Council for Freie Universitšt Berlin

ERC Consolidator Grants for Physicist Stephanie Reich and Computer Scientist Frank Noť

No 332/2017 from Nov 28, 2017

Researchers at Freie Universitšt Berlin have been awarded two prestigious grants from the European Research Council (ERC). Physicist Stephanie Reich and computer scientist Frank Noť won ERC Consolidator Grants, as the ERC announced on Tuesday in Brussels. Over five years Professor Reich will receive 2.3 million euros and Professor Noť will receive 2 million euros. Frank Noť had already received an ERC Starting Grant amounting to 1.4 million euros from 2012 to 2017, and Stephanie Reich had already had an ERC Starting Grant amounting to 1.2 million euros from 2008 to 2013. Of a total of 329 new ERC Grants approved in 22 EU member states, 56 were awarded to researchers in Germany, including these two for researchers at Freie Universitšt Berlin. The president of Freie Universitšt, Peter-Andrť Alt congratulated Professor Reich and Professor Noť, saying that this was a wonderful day for Freie Universitšt and all those involved in the winning projects and that the ERC grants signify a "huge success for the natural sciences in Berlin."

About the ERC Grant for Stephanie Reich:

With the consolidator grant, Stephanie Reich plans to develop a new field of research in which she aims to customize the reaction of materials to light. She will work with her team to develop novel measuring methods to enable the development of highly sensitive sensors and to radiate beams of concentrated sound waves.

About the ERC Grant for Frank Noť:

With the consolidator grant, Frank Noť is able to continue his excellent research, in which he and his team develop mathematical and computational physics methods in order to reveal new insights about life processes. The aim of the new ERC project is to investigate binding processes between biomolecules and to computationally simulate their dynamics on lengthscales up to that of biological cells. To facilitate this ambitious goal, new methods from the field of machine learning will be developed and employed.

Proteins move throughout the cell and are able to associate and dissociate to other biomolecules. In muscles, for example, proteins associate to long filaments that slide along each other while the muscle contracts. The force is generated by proteins of one filament that bind to those of another filament. Then they undergo a hinge motion that "pulls" the filaments relative to each other, and subsequently the proteins dissociate. Nearly all processes in human bodies are driven by such interactions between proteins that find and bind each other, exchange information, generate forces, or build up larger-scale structures.

"It is impossible to observe these processes directly," says Frank Noť. "First, proteins are extremely small - about the billionth part of a meter. Second, their geometrical structure changes very rapidly. As a result it is practically impossible to record a detailed video of processes such as protein association using microscopy or other measurements."

When simulating such processes, the main problem lies in the extreme computational cost of the simulations, explains Noť. "Even to simulate two small proteins, the interactions between about 100,000 atoms and the resulting forces need to be computed." Then, the simulation moves every atom a tiny bit into the direction of this force, while the time is advanced by a femtosecond - a tiny fraction of a second. Each of these steps is computationally expensive, but a billion times a billion steps would be needed to simulate the process of two bound proteins separating from each other.

Frank Noť’s research group and his collaborators have made ground-breaking developments that make it possible to reach these long simulation times while maintaining the dynamical information. "By combining new mathematical methods with massively parallel computations on graphics processors we could, for the first time, simulate protein-protein association and dissociation that occurs on timescales of seconds or longer with atomic resolution," says Frank Noť. This puts practically relevant applications in reach - such as the computational drug design.

"However, the long-time simulation of large protein systems, all the way up to cellular lengthscales, is currently impossible unless we discard the molecular resolution," explains Frank Noť. This is precisely the topic of the ERC project; to reach this aim, new machine learning methods will be developed and employed. If successful, this project will obtain unprecedented insights in the inner workings of the cell and enable far-reaching applications such as the battling of diseases and the optimization of biotechnological processes.