End of this page section.

Begin of page section: Contents:

Zoom into proteins

Thursday, 15 July 2021, Universität, Forschen

Researchers of the University of Graz are examining the smallest cell structures with artificial intelligence

What happens when cancer develops? What happens to the central building blocks of the cell, the proteins? Scientists can now zoom into the smallest structures and predict their make-up. An international team led by the University of Washington in Seattle (USA) has developed a new algorithm that is publicly available. Researchers of the University of Graz were thus able to decipher previously unknown protein structures with the help of the software RoseTTAfold. Now, the doors to answering biology questions more quickly are wide open, for example when combating diseases or developing drugs.

Knowledge of proteins is key: protein compounds are the central building blocks of all forms of life – from humans to bacteria and viruses. In addition, they are deeply involved in important functions such as metabolism, transport processes and signal transmission in all cells. A simple change in their structure and function can lead to the development of diseases.
The catch: Since there are hundreds of millions of proteins, the exact structure of only a small fraction is known. Deciphering them using experimental methods would be a momentous task according to Tea Pavkov-Keller from the Institute for Molecular Biosciences at the University of Graz. The development of this new algorithm drastically changes everything. “What I've been working on for years in experiments can now be predicted on the computer in a matter of hours with extreme precision and, in combination with experimental data, shows us the correct structure,” says the researcher enthusiastically.

“This accelerates research enormously”. Pavkov-Keller, a structural biologist, expects rapid development in medicine and biotechnology. Pavkov-Keller: “We can also predict the type of interaction that will occur between the individual proteins. Furthermore, RoseTTAFold records those types of proteins that consist of several so-called domains and very often form complex units.”
The ability for machine learning based on known sequences also makes it possible to predict the function and task of a protein. Based on this knowledge, connections can be changed in a targeted manner, for example to treat allergies. Tea Pavkov-Keller considers this scientific advance a quantum leap: “The effects on research and technological developments are impossible to predict at the moment.”

The ground-breaking findings will rapidly increase the pace of research in the doctoral programme Doc.funds “Biomolecular Structures and Interactions”, which is funded by the Austrian Science Fund. The training programme starts in autumn as part of BioTechMed-Graz, a cooperation between the University of Graz, the Medical University of Graz and the Graz University of Technology. Young researchers will then be working on biomolecules that play an essential role in signal transmission and the metabolism of cells.

The article was published online in the journal Science on July 16 2021.

Structure of a protein of the mushroom Phanerochaete chrysosporium, computed by RoseTTAFold.

End of this page section.

Begin of page section: Additional information:

End of this page section.