By Jonathan D. Grinstein, PhD
Chemify and Dewpoint Therapeutics have announced a partnership today to apply groundbreaking chemistry artificial intelligence (AI) to accelerate the discovery of molecules targeting cancer and neurodegeneration radically. The partnership combines Chemify’s pioneering chemistry AI and automation technology platform with Dewpoint’s condensate biology and AI technology platform.
Chemify will create a set of digitally discovered and automatically synthesized new molecules that will be used against Dewpoint’s condensate targets of interest. Dewpoint may choose to exercise an option to acquire the compounds designed by Chemify. In exchange, Chemify will receive pre-negotiated, success-driven clinical and regulatory milestones and tiered royalties.
“It was a no-brainer,” said Leroy “Lee” Cronin, PhD, CEO and founder of Chemify. Cronin is the Regius Professor of Chemistry at the University of Glasgow in Scotland, where he has been on the faculty for more than two decades.
“Dewpoint has brilliant new biological technology looking at condensates and a lot of in-house chemistry. [Dewpoint] wants to search further, deeper, and more complexly. Because Chemify is excelling at getting access to that chemical space, the idea is to get complicated molecules more quickly than they might otherwise get manually. That’s a win-win because they get more complexity quicker, and we get to validate that we can partner with a partner like Dewpoint and add value to them.”
Cronin dreams of electric sheep
Since an early age, Cronin has wanted to build a search engine to find the conditions that produced a life form from inorganic chemistry on planet Earth four billion years ago.
At some point during his research career, Cronin saw that there was no programming language for chemistry or a routine way to do any reaction reliably unless there was robotic assistance. He envisioned going from an AI-drawn chemical structure on a computer screen to making real molecules that could be used for therapeutics and more.
So he built Chemify, a company operating its proprietary molecular design, discovery, and chemical manufacturing technology to provide pharmaceutical, biotechnology, and industrial partners with better molecules. By leveraging decades of chemistry experience, hardware robotics, and artificial intelligence, Chemify’s Chemputation technology enables companies to access an end-to-end workflow for drug discovery, molecule synthesis, and material discovery.
Last month, Chemify announced funding of $43 million, including a Series A round led by Triatomic Capital, joined by new investors, including Hong Kong-based Horizon Ventures, U.S.-based Rocketship Ventures, Possible Ventures, Alix Ventures, Scotland-based Eos, and the U.K. Government Innovation Accelerators program. Existing investor BlueYard Capital also participated in the round.
With this money, Cronin has created an end-to-end molecular synthesis pipeline that consists of three modules.
Cronin and Chemify first developed a chemical programming language as a means of digitizing chemistry. With the help of state-of-the-art advances in cheminformatics, chemical engineering, and computer science, the language can make molecules that are hard to access at scale in a repeatable way. Chemify’s chemical development environment enables the robotic synthesis cluster facility to convert chemical code into actual molecules, enabling the on-demand conversion of molecule meta-representations to pure chemical products. They accomplish highly automated processes in their facility with patented, scalable, and universal chemical robotic architectures.
Secondly, Chemify is automating chemical synthesis at the lab-to-pilot plant scale to help ensure the seamless operation of their digital workflow. With their hardware platforms, complex molecules can be easily accessed, allowing for high-throughput discovery, parallel optimization, optimized reaction execution, and final product purification.
Third, using its programmable chemistry approach and artificial chemical intelligence-driven design, Chemify says it can create, triage, and synthesize bespoke libraries to discover novel molecules. They use property prediction models to design and make novel molecules with desired properties much faster and to screen large libraries to select promising candidates over design cycles, make, test, and discover. The Chemify Platform automatically translates target molecules into chemical code that can be directly executed on Chemify’s robotic synthesis platform to obtain the target molecule. Using this closed-loop approach, Chemify enables fully automated Design-Make-Test-Analyze cycles by removing the synthesis bottleneck, supercharging the discovery of novel molecules.
Cronin said there are a few different starting points for dreaming up a molecule with the Chemify platform.
“You could say, Here’s the type of molecule we like. Can you invent molecules like this and show us what’s accessible?” said Cronin. “Another one is that people can say, Here’s a physical property, like a lock and a key. Here’s a protein; make me something that binds the protein, and by the way, it needs to conduct electricity. So you can put in these different requirements, and then the search engine we have operates at different levels depending on that, and it then transmits it.”
The cool part, according to Cronin, is that while many people are getting really good at making molecules with AI, the problem is that the molecules are “locked in the metaverse,” meaning that these molecules exist theoretically only on the computer screen.
Cronin said Chemify’s answer to this pricey problem is that they lay down “railway” tracks of structures as they move through chemical space, using paths of discovery that have previously worked for them. They don’t start from a blank slate every time. So, instead of looking at every possible molecular structure, they use ones they’ve already validated in the lab.
“The brilliant thing is we have a way of checking the syntax that the program runs, and that really is literally the multi-billion dollar answer that comes out because suddenly only for sure you can make that molecule and then test it in a real cell or in whatever assay,” said Cronin. “That is where a lot of AI drug discovery materials have shown promise, but still the bottleneck is the poor human in the poor laboratory having to make this molecule that could be impossible because there’s no root to it. That’s what we’re doing at the moment.”
One type of molecule that has been on Cronin’s mind is messenger RNA (mRNA).
Cronin said, “There’s such a huge unmet need in formulating mRNA properly with the right lipids and molecules because RNA is not just RNA—it’s all messed up. There’s iron everywhere; it’s dirty, and it’s folded wrong. I thought, ‘I can probably refold all of this using this little engine,’ but that’s another product release later in the year.”
Where the rubber meets the road
While Cronin is excited to put Chemify’s approach to the test in their partnership with Dewpoint, he admits that the pressure is on.
“It’s a great chance to do something good, but every time, I keep reminding myself that drug discovery is a bit like playing roulette in Las Vegas—as long as you don’t get caught for cheating and you faithfully spin the wheel and put the ball in,” said Cronin. “We’re doing a lot more than that, and I expect that we will generate some interesting outcomes for [Dewpoint].”
There’s the allure of looking novel, and then there’s demonstrating by collaborating with partners and working within markets.
Revealed to be a bit of a sci-fi nerd, Cronin likens his approach to the title of Philip K. Dick’s classic book, where, instead of sheep, he dreams of electric drugs. Cronin isn’t here to sell his dreams. He’s here to sell his reality. This first partnership will put Cronin and Chemify to the test of being able to “make the ‘meta’ physical.”