Home https://server7.kproxy.com/servlet/redirect.srv/sruj/smyrwpoii/p2/ Science https://server7.kproxy.com/servlet/redirect.srv/sruj/smyrwpoii/p2/ DeepMind solves protein folding ‘grand challenge’ with AlphaFold AI

DeepMind solves protein folding ‘grand challenge’ with AlphaFold AI



Demis Hassabis, CEO of Alphabet, the Google DeepMind research team, at Google’s Future of Go Summit in China on May 23, 2017.

LONDON – Alphabet-owned DeepMind has developed a piece of artificial intelligence software that can accurately predict the structure that proteins will unfold in a matter of days, and solve a 50-year “big challenge”

; that could pave the way for a better understanding of diseases and drug discovery.

Each living cell has thousands of different proteins inside that keep it alive and well. It is important to predict the form that a protein will fold into because it determines their function, and almost all diseases, including cancer and dementia, are related to how proteins work.

“Proteins are the most beautiful, most beautiful structures, and the ability to predict exactly how they fold together is really, very challenging and has occupied many people for many years,” Professor Dame Janet Thornton of the European Bioinformatics Institute told reporters during a calls.

The British research laboratory DeepMind’s “AlphaFold” AI system participated in a competition organized by a group called CASP (Critical Assessment for Structure Prediction). It is a social experiment organization tasked with speeding up solutions to a problem: how to calculate the 3D structure of protein molecules.

CASP, which has been monitoring progress in the field for 25 years, compares competition submissions to an “experimental gold standard.” On Monday, it said DeepMind’s AlphaFold system has achieved unparalleled levels of accuracy in protein structure prediction.

“DeepMind has jumped ahead,” said Professor John Moult, president of CASP, in a press release prior to the announcement. “A 50-year major challenge in computer science has largely been solved.”

Moult added that there are “major influences a bit down the line for drug design” and within the emerging field of protein design.

With around 1,000 employees and almost no revenue, DeepMind has become an expensive company that Alphabet (Google’s parent company) can support. However, it has emerged as one of the leaders in the global AI race along with Facebook AI Research, Microsoft and OpenAI.

The breakthrough was welcomed by Google CEO Sundar Pichai on Twitter.

DeepMind co-founder and CEO Demis Hassabis said in the call: “The ultimate vision behind DeepMind has always been to build general AI and then use it to help us understand the world around us by greatly accelerating the pace of science. discovery. ”

The company, which Google bought for $ 600 million in 2014, is best known for creating AI systems that can play games like Space Invaders and the ancient Chinese board game Go. However, it has always said that it will have more scientific impact.

“Games are good evidence of effectively developing and testing general algorithms that we once hoped we would transfer to real-world domains as scientific problems,” Hassabis said. “We feel AlphaFold is a first proof of this dissertation. These algorithms are now becoming mature enough and strong enough to be applicable to truly challenging scientific problems.”

DeepMind also participated in a CASP protein folding competition in 2018. While the results at the time were impressive, John Jumper, AlphaFold manager at DeepMind, said the team knew it was somehow producing something with “really strong biological relevance or be competitive with the experiment. “

This year’s competition, however, was not ordinary sailing, and Jumper said that DeepMind went for three months without making any progress. “We would sit there and worry, have we exhausted the data?” he said.

Even as the deadline for the competition approached, Jumper and his team were still worried that they might have made a mistake. “There could always be a flaw creeping into machine learning systems,” he said.

But their efforts seem to have paid off. “We truly believe we have built a system that provides accurate and actionable information to experimental biologists,” he said. “The reason you have a structure is to understand something about the natural world and then ask even more questions. We think we’ve built a system that really helps people do that.”


Source link