
Gary Glick puts Odyssey's $218M cash stack to use, acquiring machine learning outfit
Around the same time serial entrepreneur Gary Glick was putting together his latest (and biggest yet) venture, Odyssey Therapeutics this past March, a mentee introduced him to a young London-based company working on applying machine learning to drug discovery.
Rahko, founded just three years ago by a few machine learning experts, was developing a platform right up Glick’s alley: Odyssey, as he’s conceived it, would execute on drug discovery at top speed just like IFM and Scorpion, his previous startups, but do it with a heavy dose of data science.
The initial idea was to team up and leverage Rahko’s molecular simulation and machine learning capabilities to design compounds for Odyssey.
But it turned into something much deeper.
Odyssey, flush with $218 million in venture dollars, revealed Thursday morning that it will acquire Rahko, integrating its 13 staffers into a 100-plus team — while adding more people to the team, which will remain in London.
With some of the “leading minds” in the area of generative modeling, Glick noted that Rahko could turbocharge Odyssey’s R&D efforts in one of two ways: approaching targets it otherwise couldn’t, and also coming up with better molecules than it otherwise could create.
“If you don’t have a hit, there’s no program,” Glick said. “That’s a big bottleneck. And that’s not insignificant. There are many, many examples in pharma where people run high throughput screens for a target of interest that just don’t get anything.”
As for what those R&D efforts are about exactly, Odyssey is remaining secretive, divulging only that it’s working on inflammation and oncology — and building on “past approaches” of anti-TNF antibodies, JAK inhibitors and targeted cancer immunotherapies.
Glick much preferred talking about how Odyssey’s internal processes set it apart from traditional drugmakers — which he believes has been crucial to recruiting people like Heather Carlson, an expert in computational drug discovery who’s leaving the University of Michigan to join the biotech.
Typically, in the industry, computational chemistry is seen as a support service. And so, some program or groups of programs may send structures to a computational chemist, and they’ll say, just tell me the best one. And, you know, they do what they do, and they ship them back. And a lot of them find that kind of unrewarding […] What we have done differently, certainly to many large pharma, is that you know, whether you’re a synthetic organic chemist that’s been practicing medicinal chemistry, or whether you’re a computational chemist, or a data scientist that’s been practicing drug hunting, they are both part of a team, they both get to design compounds, those compounds are made. Project teams are resourced with appropriate — appropriately sized with chemistry resources that all the compounds can be made. That’s how you get the best decisions.