Hand in hand with artificial intelligence, DeepMind has created the most comprehensive map of human proteins

The company DeepMind -which was acquired in 2016 by Alphabet, parent company of Google– and the European Molecular Biology Laboratory (EMBL) used the artificial intelligence system AlphaFold publish the most complete and accurate database of predictions of the structures of the human proteins.

The database, open to the scientific community and which will be hosted by the European Institute of Bioinformatics (EMBL-EBI), will include approximately 20,000 proteins expressed by the human genome.

Among the first 350,000 structures published In the database, in addition to the human proteome, are the proteins of 20 biologically important organisms such as E. coli, fruit fly, mouse, zebrafish, malaria parasite and tuberculosis bacteria.

With this, the accumulated knowledge about the structures of proteins is significantly expanded, more than double the number of human protein structures with high precision predictions available to researchers, accelerating work in a wide variety of fields, EMBL noted in a note.

The magazine Nature published today a study describing how these predictions are made and provides the most complete picture of the proteins that make up the human proteome (the set of proteins encoded by the human genome), the understanding of which is of great importance for health and medicine.

The resulting data set provides a reliable prediction of the structural position of nearly 60% of the amino acids of the human proteome.

The authors of the study found that the algorithm AlphaFold was able to predict “with confidence” the structural position of 58% of the amino acids of the human proteome.

Of these, the position of a 35.7% subset was predicted with a degree of confidence “very high“, Which means twice the number covered by experimental structures, explained the magazine.

Proteins have a unique three-dimensional shape that makes them fit together, but figuring out this is a big challenge. The use of artificial intelligence has made it possible to create the most comprehensive database of predictions of how these molecules fold up.

The structure of each protein – a fundamental element of life -, which depends on the amino acids that compose it, defines what it does and how it does it. Therefore, being able to determine it provides valuable information for understanding biological processes, with the aim of advancing in various fields of research and for drug development in the future.

Researchers believe that accurately predicting large-scale structures will become “in an important tool which will allow to approach new scientific questions from a structural point of view», And the predictions of AlphaFold will help clarify further the role of proteins.

“We believe this is the most important contribution artificial intelligence has made to the advancement of scientific knowledge to date., and this is a great example of the types of benefits that artificial intelligence can bring to society, ”according to the founder of DeepMind, Demis Hassabis.

The use of artificial intelligence, with its ability to predict by calculation The shape of a protein from its amino acid sequence does not have to be determined experimentally with the use of laborious and sometimes expensive techniques.

AlphaFold was formed with data from public resources created by the scientific community, it is therefore logical that their predictions are public, defended the director general of the EMBL, Edith Herad.

This tool, which for Herad is “A real revolution for the life sciences, just as genomics was decades ago, ”is already being used by the Drugs for Neglected Diseases Initiative.

In addition, a group from the University of California in San Francisco used the predictions of this algorithm to study the biology of SARS-CoV-2.

(With information from EFE)

Read on: