Alphabet kickstarts new AI drug discovery outfit with lessons learned from DeepMind protein breakthroughs
Last summer marked a major breakthrough in drug discovery when DeepMind, a predictive modeling startup from Google parent company Alphabet, offered the most accurate picture yet of the “protein folding” problem. The Alphabet team is now propping up a unit focused solely on drug discovery, and it will look to leverage lessons learned from DeepMind’s example.
Alphabet has launched Isomorphic Labs, a London-based drug discovery startup leveraging the company’s AI and machine learning work, and lessons from DeepMind’s AlphaFold breakthroughs, CEO Demis Hassabis said in a blog post Thursday.
You’ll know Hassabis as the CEO of Alphabet’s DeepMind, the company that offered the most accurate visual model yet of “protein folding,” a notorious challenge that had eluded researchers for more than 50 years.
Hassabis will remain in his role at DeepMind while leading what he called the “first phase” of growth at Isomorphic, including fleshing out the company’s strategic vision and hiring on key staff. Here’s what he had to say about the company’s mission:
We are at an exciting moment in history now where (AI and machine learning) are becoming powerful and sophisticated enough to be applied to real-world problems including scientific discovery itself. One of the most important applications of AI that I can think of is in the field of biological and medical research, and it is an area I have been passionate about addressing for many years. Now the time is right to push this forward at pace, and with the dedicated focus and resources that Isomorphic Labs will bring.
While there will be some crossover in terms of Isomorphic and DeepMind’s expertise, a company spokesperson said the units would operate independently with the former focusing on creating a “computational platform to understand biological systems from first principles to discover new ways to treat disease.”
Isomorphic is, of course, not the first company looking to leverage breakthroughs in AI to change the game of drug development, a chronically tricky endeavor with very low rates of success. Companies in the vein of insitro and Exscientia have claimed ownership of “AI-discovered” molecules — with some debate over who finished first — but those recent results still have a long way to go to prove themselves in the clinic.
But coming from the team behind AlphaFold, Isomorphic is likely to catch a lot of eyeballs as a relatively late entrant into the field. In July, DeepMind announced it would open its protein database, culled from AlphaFold, to science in what looked like a major win for drug developers. AlphaFold’s predictive modeling system uses protein sequences to offer visual models for researchers, potentially shedding new light on drug targets, experts say.
While there appears to be some crossover in terms of what Isomorphic and DeepMind are doing on this front, Alphabet said AlphaFold will still be working in this space separately, with potential learnings shared across groups. That’s a good thing because actually making meaningful headway in drug discovery will likely take as many shots on goal as possible.
“Biology is likely far too complex and messy to ever be encapsulated as a simple set of neat mathematical equations. But just as mathematics turned out to be the right description language for physics, biology may turn out to be the perfect type of regime for the application of AI,” he wrote.