Isomorphic Labs is Alphabet’s game in AI drug discovery – TechCrunch
The field of drug discovery has been supercharged by the capabilities of AI, which several companies have applied in various ways to turn a huge practical problem into a manageable information problem. The latest to do so is Alphabet, the parent company of Google, which created Isomorphic Labs, under the leadership of DeepMind, Demis Hassabis, to try its luck in this promising new field.
Very little was revealed about the company in its first blog post and a very general FAQ that accompanies it. The company’s goal is to “build an IT platform to understand biological systems from the very first
principles to discover new ways of treating disease.
There are, of course, a few assumptions built into this foundational statement, including that it is possible to computer simulate biological systems in a field conducive to drug discovery.
Several large companies have been formed and funded with hundreds of millions of dollars to pursue very similar goals over the past five years or so, and there has not been any visible revolution or famous miracle drug discovered by AI for one. previously incurable disease. The reasons why are beyond the scope of this article (and will no doubt be addressed by Isomorphic Labs in the near future), but it is clear that these AI systems are not miracle factories, just miracle factories. parts of a long and complex process that always involves a lot of time, money and test tubes.
Hassabis is no fool, and although he describes biology rather optimistically as “an information processing system, albeit extraordinarily complex and dynamic” (I can sense readers in this field scrolling down to comments now), he tempers this shortly after:
Biology is probably far too complex and messy to ever be encapsulated as a simple set of neat mathematical equations. But just as mathematics has turned out to be the right descriptive language for physics, biology may turn out to be the perfect type of diet for the application of AI.
The idea that information systems and biological systems can have a common structure is the inspiration for the name, isomorphic systems; isomorphic means of the same shape but having a different origin.
His reasoning undoubtedly comes in part from the effectiveness of DeepMind’s AlphaFold, an AI-powered protein folding system that blew up biologists last year and helped create a new normal in a field. very complex.
DeepMind’s learning systems have shown a particular affinity for generality or knowledge transfer, i.e. having a structure capable of being reused for very different tasks. And if, as AlphaFold’s success suggests, biological systems fit well with this kind of simulation and analysis, Hassabis’s assessment of the company’s broader capabilities may be true.
If so, it won’t be for a while. Even with the ongoing startup provided by DeepMind’s AI research (which will remain separate but can be shared), Isomorphic is essentially starting from scratch on this issue. It’s about hiring a “world-class multidisciplinary team” and maybe in a year or two we’ll see the first signs of results emanate from the company’s ambitions.