[News] An artificial intelligence from Google unravels a million-year-old mystery in minutes

[MosterOfficial ☆]

 

For more than half a century, humanity has faced the challenge of finding out the form of the basic building blocks of life, a knowledge that would help cure diseases that today have no remedy. In 1969 a biologist predicted that it would take more than 14 billion years to figure out the possible configurations of a single protein, but the AlphaFold system has managed to do it in just a few minutes. The "instructions" of more than 20,000 types of proteins that are responsible for replicating genetic material, obtaining energy or producing antibodies, among other functions, are stored in human DNA. Biologists try to understand their structures in processes that take months or years, but this artificial intelligence system that belongs to Google is capable of doing it in a few minutes. This system, called AlphaFold predicts the structure of proteins after reading the long chains of molecules that compose it, it is as if you could predict the appearance of a recipe just by reading the ingredients. DeepMind, the company that created this system, and the European Molecular Biology Laboratory published this Thursday in the journal 'Nature' more than 350,000 protein structures.

 

An unprecedented advance in biology

 

The scientist Venki Ramakrishnan, winner of the Nobel Prize in Chemistry in 2009, affirms, as collected by ‘El País’, that it is “an amazing advance”, with unpredictable consequences. “It has happened much earlier than many experts would have predicted. It is going to be exciting to see the many ways in which biological research is going to radically change, "said the scientist in a statement. These structures could be used to investigate treatments for diseases such as Alzheimer's, cancer and even COVID, since all of them are related to the form of some protein. The director of the company, Demis Hassabis announced that in the coming months he will publish 100 million structures calculated with this AlphaFold system.

 

Prediction with errors

 

The Spanish biologist, José Antonio Márquez has explained to 'El País' that elucidating the shape of a protein with a synchrotron, or with the alternative method of electronic cryomicroscopy, may require “months or even years”, while this novel system done in a few minutes, albeit with errors. “These are computer predictions, not the experimental determination of the structure. And the precision is 58% ”explains the biologist. The scientist assures that, although it is not exact, this system will save a lot of time in research. In addition, he also points out that there is another limitation, since proteins tend to interact with others, and what AlphaFold does is predict isolated molecules, so it is not capable of predicting interactions between cells. "It is going to accelerate discoveries in practically all areas of biology," says Márquez.

Biology