Academics and researchers are debating claims by the Google-owned artificial-intelligence firm DeepMind that it has solved one of the toughest problems in biology.
On Monday, DeepMind said it had broken new ground in understanding the behavior of microscopic proteins. It said its AlphaFold artificial-intelligence program could reliably predict their shape, which would effectively solve a problem that's plagued scientists for decades.
Professor Venki Ramakrishnan, a winner of the Nobel Prize in chemistry, hailed the results as a "stunning" achievement.
And DeepMind's team wrote in a blog on Monday: "This breakthrough demonstrates the impact AI can have on scientific discovery and its potential to dramatically accelerate progress in some of the most fundamental fields that explain and shape our world."
But some academics are skeptical of how much DeepMind's claims should be hyped as a "solution" to the protein-folding problem. They called on DeepMind to put AlphaFold's code into the public domain and said it wasn't clear how the program would perform outside a narrow setting.
DeepMind's breakthrough was part of the Critical Assessment of Structure Prediction (CASP), a global competition specifically set up to test research teams on their ability to predict a protein's shape from its sequence of amino acids.
Max Little, an associate professor and senior lecturer in computer science at the University of Birmingham, told Business Insider DeepMind's AI had shown potential only "within the context of the CASP database challenge".
He said: "We can't really be sure how well AlphaFold will work when faced with the far more rich and varied array of proteins found in the real world of living organisms."
Here's what DeepMind did
Proteins are a key building block for life, found in humans, animals, plants, and microscopic organisms. They are invisible to the human eye and constantly rearranging themselves, which makes studying and predicting their behavior hard.
The way proteins move around (or "fold") inside people's bodies — transforming from a string of amino acids into more complicated 3D structures — has big implications for their health and is linked to everything from Alzheimer's disease to the flu. That's why scientists have spent the better part of 50 years trying to predict their movement.
If scientists knew how a protein would behave, experts say, they could theoretically alter its behavior. For example, if a misfolding protein is stopped in its tracks, its host could be saved from contracting a neurodegenerative disease like Parkinson's. Medical personnel could also better target treatment because they would have a better idea of how a person's body would react, staving off any nasty side effects in advance.
In 2018, DeepMind first entered AlphaFold into CASP, but the results weren't deemed concrete enough to be medically useful.
This year, the latest version of AlphaFold had been trained on a "protein data bank" made up of about 170,000 structures and matched up with predictions made by scientists in labs — a much longer and more expensive process — with high accuracy in two-thirds of cases.
Experts say DeepMind's research might not apply outside a narrow setting
Professor Michael Thompson, an expert in structural biology at the University of California, Merced, branded the idea that protein folding had been solved "laughable."
"Frankly, the hype serves no one,"he wrote on Twitter, adding that the company could "never live up to the promise that's been made."
He then said: "Until DeepMind shares their code, nobody in the field cares and it's just them patting themselves on the back."
Thompson did say "the advance in prediction is impressive." He added: "However, making a big step forward is not the same as 'solving' a decades old problem in biology and chemical physics."
"Despite this being the biggest thing that has happened in protein folding, the problem is not yet solved," Vishal Gulati, a seasoned venture capitalist specializing in deep tech and health startups, said in a blog post published on Tuesday.
He said DeepMind's research was "not a minor achievement" but added: "Compared to the problem of protein folding, CASP is a game. It is a very hard game but it is a reduced problem set which helps us train tools and standardize performance ... It is a necessary step but it is not sufficient."
Lior Pachter, a professor of computational biology at the California Institute of Technology, agreed, writing on Twitter that the protein-folding problem was not "well defined."
"I don't mind that Google hyped this," he wrote. "I do mind that many (computational) biologists who ought to know better are going around screaming: 'Protein folding is solved!'
"Have some self-respect."
But CASP's chair said he had looked at the results and that code sharing wasn't the norm
That the findings would trigger some robust public debate is, perhaps, to be expected.
In an email exchange with Business Insider, CASP Chair John Moult rejected the criticisms, writing: "CASP is not a game, it's a scientific experiment designed to test folding methods in close-to-real-life situations ... What is missing?"
He added: "Perhaps the people hyping the results have actually looked at them? Seems like this person hasn't."
In response to criticism of DeepMind's refusal to share AlphaFold's code, Moult wrote: "This is an old chestnut in the field. Although code-sharing is obviously desirable, and some groups do it, it has never been the norm. Not clear why DeepMind should be held to a higher standard than others."
He then said: "Fifty years of listening to false claims about this problem has made me the world's biggest skeptic. But I have looked at these results very carefully ... Clearly, this is just the beginning of what DeepMind and others will achieve with these sorts of approaches."
A spokesperson for DeepMind declined to comment when approached by Business Insider.
Join the conversation about this story »
NOW WATCH: July 15 is Tax Day — here's what it's like to do your own taxes for the very first time