The expanse of disease-causing proteins is fertile hunting ground for drug discovery. Many biotech startups are trying to find small molecules capable of binding to these targets. That’s great when it works, but many protein targets remain elusive, said Raphael Townsend, founder and CEO of Atomic AI. Instead of going after disease-causing proteins, Townsend’s startup is aiming for a different target: the RNA that carries instructions for making those proteins. Atomic AI uses artificial intelligence to find ways to drug RNA and is now out of the stealth, backed by $35 million.
“By targeting the RNA instead, you give yourself new ways to attack these untreatable diseases,” Townsend said.
The Series A funding round announced Wednesday was led by Playground Global.
In order to administer RNA as a drug, scientists must first understand it better. Proteins are relatively well understood, with hundreds of thousands of known protein structures, Townsend said. In comparison, the human transcriptome, the complete set of all RNA, is poorly understood. The hundreds of known RNA structures are less well mapped compared to proteins, Townsend said. That’s key because there’s growing recognition that RNA alone plays an important role in disease, he added.
Proteins fold and change shape, which can make them difficult to target with a small molecule. But RNA is far more flexible, making it more of a moving target, Townsend said. San Francisco-based Atomic AI’s technology maps the transcriptome using an approach that combines wet lab experiments with computational analysis. The data generated by the wet lab will be used to train the AI to discover new targets on the three-dimensional structure of RNA, Townsend said. The AI makes predictions that inform additional wet lab experiments. These results feed additional AI analysis and continue a virtuous cycle.
Atomic AI’s technology is based on research from Townsend’s PhD thesis at Stanford University. This research was published in the journal Science in 2021, the same year Atomic AI was founded. Since then, the company has made strides with its algorithms and wet lab, Townsend said. The technology, now called Platform for AI-driven RNA Structure Exploration (PARSE), has also improved in terms of speed and accuracy.
The new capital allows Atomic AI to scale the platform, which could help the startup grow into a drug discovery organization, Townsend said. The company will begin to narrow down the goals to be pursued. Townsend declined to identify specific diseases that Atomic AI could track, but said the technology could be used to discover small molecules for use in oncology, neurodegenerative diseases, cardiology, rare diseases and infectious diseases. The startup’s initial research will focus on identifying the parts of the transcriptome that are targeted in the first place, Townsend said.
Atomic AI isn’t the first biotech to target drugs using RNA, and in addition to an earlier launch, some of these startups already have partnerships with big pharma. Arrakis Therapeutics’ most advanced program is an oncology lead optimization compound. Based in Waltham, Massachusetts, the company has a drug discovery alliance with Roche. Skyhawk Therapeutics is another Waltham-based company developing small molecules that target RNA. This company has alliances with Bristol Myers Squibb, Merck and Takeda Pharmaceutical. Rather than targeting RNA directly, Remix Therapeutics is developing drugs that target parts of the cell that process them. Nearly a year ago, Cambridge, Massachusetts-based Remix formed a research alliance with a Johnson & Johnson subsidiary. Arpeggio Biosciences, based in Boulder, Colorado, recently announced a $17 million Series A funding round.
Townsend acknowledges the other companies pursuing RNA-targeted small molecules, but he says what sets Atomic AI apart is the wet-lab component of its platform. Companies that take an AI-only approach to RNA will struggle because there just isn’t a lot of RNA data for those technologies to analyze, he explained.
Now that Atomic AI is no longer secretive, Townsend said he’s on the lookout for potential partnerships. While the startup’s internal research will focus on developing small molecule drugs, Townsend said partnerships will focus on using PARSE to develop new RNA-based drugs. The platform’s ability to predict how RNA folds and forms new structures can be used to develop new RNA drugs, he explained. The technology also has potential to improve certain aspects of RNA-based drugs, such as stability. For example, a more stable RNA molecule could avoid the ultra-cold storage required for the messenger RNA-based Covid-19 vaccines.
Atomic AI initially raised $7 million in 2021 seed funding led by 8VC. This company also invested in the Series A round, which included participation from the factory’s headquarters; gray lock; Not boring; AME Cloud Ventures; and angel investors, including former GitHub CEO Nat Friedman; Doug Mohr; Curai CEO Neal Khosla; and Patrick Hsu, professor at the University of California, Berkeley, and co-founder of the Arc Institute.
Image by libre de droit, Getty Images