Three-dimensional structure of a protein (illustration)
Photo: Christoph Burgstedt / Science Photo Library / imago images
Not a single cell in our body can do without them: Proteins are essential building blocks of living organisms.
Your blueprint is laid down in DNA.
Proteins include the components of the immune system, such as antibodies, as well as hormones such as insulin or enzymes that break down lactose in the small intestine and make milk compatible.
Exactly how a protein works in the body depends not only on the order of the atoms it consists of - but also on its structure: the long chains of amino acids that make up proteins twist into complex 3D structures.
And their decryption is often considered tedious and time-consuming;
The method of X-ray crystal structure analysis is mainly used.
At the end of last year, researchers at the British company DeepMind, a subsidiary of the Google holding company Alphabet, reported that a novel approach had been successful.
An artificial intelligence (AI) developed by them called AlphaFold can therefore precisely predict the structures of molecules.
Now the team around Kathryn Tunyasuvunakool, John Jumper and Demis Hassabis from DeepMind has presented the application of the method in a specialist article in the magazine "Nature".
The result is a freely accessible database that contains many thousands of protein structures.
These are to be used for breakthroughs in medical research, but also for plant breeding - or for the development of bacteria that can break down plastic in the environment.
Almost the entire amount of all 20,000 proteins occurring in humans, the so-called proteome, is covered, but with different levels of accuracy.
In addition, there are tens of thousands of other proteins that play a role in other model organisms that are important for research, such as mice, fruit flies and coli bacteria.
Knowing the structure is so important because it can be used to predict how other molecules will bind to the protein - i.e. what effect a certain substance will have in practice.
"We believe that this is the most complete and accurate image of the human proteome to date," says Hassabis, who is also the head of DeepMind.
His company is also of the opinion "that this work represents the most significant contribution that Artificial Intelligence has made to the advancement of scientific knowledge".
"This will revolutionize our understanding of how life works," says Edith Heard from the European Molecular Biology Laboratory (EMBL) in Heidelberg.
"The possible applications are only limited by our imagination." The database of protein structures will be maintained at EMBL in the future.
And Gira Bhabha, who studies cell biology at New York University, praises the time saved for projects in various areas of research: “Regardless of whether you are studying neuroscience or immunology - whatever your field of biology is - this can be useful. «
chs / jme