Happy Nice Mall

John McGeehan, biologist and director of the Center for Enzyme Innovation in Portsmouth, England, has been searching for a molecule that can break down the 150 million tons of soda bottles and other plastic waste that have been scattered around the world for several years.

Working with researchers on both sides of the Atlantic, few good options. But his job is the job of the most meticulous locksmith: to identify the chemical compounds that can bend and fold on their own to fit snugly into the molecules of a plastic bottle and separate them like a key to a door. .

Determining the exact chemical content of any enzyme is a fairly simple challenge these days. But determining its three-dimensional shape may require years of biochemical experiments. Last fall, after reading a call from an artificial intelligence lab in London DeepMind had set up a system that automatically predicted the shapes of enzymes and other proteins., Dr. McGeehan asked the lab if he could help with his project.

Toward the end of a workweek, he sent DeepMind a list of seven enzymes. The next Monday, the lab gave figures for seven of them. Dr. “This got us one year ahead, if not two years from where we are,” McGeehan said.

Now, any biochemist can speed up their work in pretty much the same way. DeepMind on Thursday published predicted shapes of more than 350,000 proteins, the microscopic mechanisms that drive the behavior of bacteria, viruses, the human body and all other living things. This new database includes three-dimensional structures of all proteins expressed by the human genome, as well as proteins seen in 20 other organisms, including mice, fruit flies, and E. coli bacteria.

Providing nearly 250,000 previously unknown shapes, this vast and detailed biological map could accelerate the ability to understand diseases, develop new drugs, and reuse existing drugs. It could also lead to new kinds of biological tools, such as an enzyme that efficiently breaks down plastic bottles and turns them into materials that can be easily reused and recycled.

“This can get you ahead in time — it can affect the way you think about problems and help you solve them faster,” said Gira Bhabha, assistant professor in the department of cell biology at New York University. “Whether you’re studying neuroscience or immunology – whatever your field of biology – it can be useful.”

This new information is a key in itself: If scientists can determine the shape of a protein, they can also determine how other molecules will bind to it. This can reveal, for example, how bacteria resist antibiotics and how to counter that resistance. Bacteria show resistance to antibiotics by expressing certain proteins; If scientists can identify the shapes of these proteins, they can develop new antibiotics or new drugs that suppress them.

In the past, detecting the shape of a protein required months, years, or even decades of trial-and-error experiments on the lab bench involving X-rays, microscopes, and other tools. But DeepMind can significantly shrink its timeline with its AI technology known as AlphaFold.

Dr. When McGeehan sent DeepMind his list of seven enzymes, he told the lab he had identified shapes for two of them, but did not say which. This was a way to test how well the system was working; AlphaFold passed the test and correctly predicted both shapes.

Dr. McGeehan said it was even more remarkable that the estimates arrived in a matter of days. He later learned that AlphaFold actually completed the task in just a few hours.

AlphaFold predicts protein structures using what’s called “a”. plexusA mathematical system that can learn tasks by analyzing large amounts of data (in this case, thousands of known proteins and their physical shapes) and making predictions into the unknown.

It’s the same technology identifies commands you bark at your smartphone, Recognizes faces in photos you share on Facebook and that translates one language to another On Google Translate and other services. However, many experts believe that AlphaFold is one of the most powerful applications of the technology.

“This shows that AI can do useful things within the complexity of the real world,” said Jack Clark, co-author of the AI ​​Index, an effort to track the progress of artificial intelligence technology around the world.

Dr. As McGeehan discovered, it can be extremely accurate. AlphaFold can predict the shape of a protein about 63 percent of the time with an accuracy that rivals physical experiments, according to independent benchmark tests that compare its predictions with known protein structures. Most experts had assumed that such a powerful technology was still years away.

“I thought it would take another 10 years,” said Randy Read, a professor at Cambridge University. “It was a complete change.”

However, the accuracy of the system varies, so some estimates in DeepMind’s database will be less useful than others. Every prediction in the database comes with a “confidence score” that shows how accurate it could be. DeepMind researchers estimate that the system provides a “good” estimate about 95 percent of the time.

As a result, the system cannot completely replace physical experiments. Used alongside working on the lab bench, it helps scientists determine what experiments they should run and fills in the blanks when experiments fail. Using AlphaFold, researchers at the University of Colorado Boulder have recently helped identify a protein structure they’ve been struggling to identify for more than a decade.

The developers of DeepMind have chosen to freely share the database of protein structures rather than selling access, in hopes of promoting progress in the biological sciences. “We’re interested in maximum impact,” said Demis Hassabis, CEO and co-founder of DeepMind, which is owned by the same parent company as Google but operates more like a research lab than a commercial enterprise.

Some scientists compared DeepMind’s new database to the Human Genome Project. The Human Genome Project, completed in 2003, provided a map of all human genes. Now, DeepMind has provided a map of the roughly 20,000 proteins expressed by the human genome – another step towards understanding how our bodies work and how we can react when things go wrong.

The hope is that technology continues to evolve. A lab at the University of Washington built a similar system called RoseTTAFold and openly shared the computer code that powered its system, like DeepMind. Anyone can use technology and everyone can work to improve it.

Even before we started sharing DeepMind technology and data openly, AlphaFold nurtured a wide variety of projects. University of Colorado researchers are using technology to understand how bacteria such as E. coli and salmonella develop resistance to antibiotics and to develop ways to combat that resistance. Researchers at the University of California at San Francisco used the tool to improve their understanding of the coronavirus.

The coronavirus harms the body through 26 different proteins. With the help of AlphaFold, the researchers developed the understanding of a key protein and they hope the technology can help them better understand the other 25.

If this comes too late to make an impact on the current pandemic, it can help prepare for the next one. “A better understanding of these proteins will help us target not just this virus but other viruses as well,” said Kliment Verba, one of the researchers in San Francisco.

The possibilities are innumerable. DeepMind, Dr. After giving McGeehan figures for seven enzymes that could rid the world of plastic waste, he sent 93 more lists to the lab. “They’re working on those now,” he said.