[ad_1]
A artificial intelligence The company that gained fame for designing computer systems that can beat humans in games has now made huge strides in biological science.
The company DeepMind, which is owned by the same parent company as Google, has created an artificial intelligence system that can quickly and accurately predict how proteins fold into their 3D shapes, a surprisingly complex problem that has plagued researchers for decades. decades, according to The New York Times.
Determining the structure of a protein can require years or even decades of painstaking experimentation, and current computer simulations of protein folding lack precision. But DeepMind’s system, known as AlphaFold, took only a few hours to accurately predict a protein’s structure, the Times reported.
Related: Why does artificial intelligence scare us so much?
Proteins are large molecules essential for life. They are made up of a chain of chemical compounds called amino acids. These “chains” fold together in complex ways to create unique structures that determine what the protein can do. (For example, the spike protein on the novel coronavirus allows the virus to bind and invade human cells.)
Nearly 50 years ago, scientists hypothesized that the structure of a protein could be predicted by knowing only its amino acid sequence. But solving this “protein folding problem” has proven to be extremely difficult because there are a bewildering number of ways the same protein could theoretically fold into a 3D structure, according to a statement from DeepMind.
Twenty-five years ago, scientists created an international competition to compare various methods of predicting the structure of proteins – something of an “Olympic of proteins,” known as CASP, which stands for Critical Assessment of Protein Structure Prediction, according to The Guardian.
In this year’s challenge, AlphaFold’s performance was above that of its competitors. It achieved a level of precision that researchers hadn’t expected to see in years.
“This computational work represents a staggering breakthrough on the problem of protein folding, a great 50-year challenge in biology,” Venki Ramakrishnan, President of the Royal Society in the UK, who was not involved in this work, said in a press release. “This happened decades before many people in the field predicted it. It will be exciting to see the many ways this will fundamentally change biological research.”
For the competition, the teams receive the amino acid sequences of around a hundred proteins, the structures of which are known but not published, according to Nature news. Predictions are scored from zero to 100, with 90 being considered equivalent to the accuracy of the experimental methods.
AlphaFold trained to recognize the relationship between amino acid sequence and protein structure using existing databases. Then, he used a neural network – a computer algorithm modeled on how the human brain processes information – to iteratively improve his prediction of unpublished protein structures.
Overall, AlphaFold had a median score of 92.5. This represents a score of less than 60 that the system achieved in its first CASP competition in 2018.
The system isn’t perfect – in particular, AlphaFold has not performed well in modeling groups of proteins that interact with each other, Nature News reported.
But this advance is a game-changer.
“I think it’s fair to say that it will be very disruptive to the field of protein structure prediction. I suspect that many will leave the field as the central problem has arguably been solved,” said Mohammed AlQuraishi, computer biologist at Columbia University. . “This is a first-rate breakthrough, certainly one of the most significant scientific findings of my life.”
DeepMind has already made headlines by creating an AI program, known as AlphaGo, which beat humans at the old game of Go.
The researchers hope that AlphaFold can have many applications in the real world. For example, it could help identify the structures of proteins involved in certain diseases and speed up drug development.
DeepMind is currently working on a peer-reviewed article on its work on AlphaFold, the Times reported.
Originally posted on Live Science.
[ad_2]
Source link