Exclusive: Microsoft, Sam Altman back a new AI biotech upstart
Updated: Apr 1
Jason Mast, EndPointsNews, March 23, 2021 10:01 AM EDT,
Jen Nwankwo, 1910 Genetics CEO
Most artificial intelligence biotechs start with a computer scientist or two and an algorithm. Jen Nwankwo started from the other side of the spectrum.
She had just gotten her PhD from Tufts University in 2016, a dyed-in-the-wool pharmacologist who had received an HHMI fellowship and worked on sickle cell drug discovery at Boston Children’s Hospital, and was working at Bain Capital when she started reading up on artificial intelligence. She’d pour over every news article she saw on self-driving cars or image recognition, wondering with each word how she could apply the same technology to the problems that plagued her as a drug developer.
“I don’t come from the technology world, I am not myself an AI person,” Nwankwo told Endpoints News. “I’m what you call an AI enthusiast.”
The lack of tech experience, though, hasn’t cost her with tech investors. After building a platform and launching a company, 1910 Genetics, around those early musings, she has convinced Microsoft’s VC firm M12 and the deep tech VC Playground Global to back a $22 million Series A, with the goal of turning the platform she and a colleague built into a long list of drug development programs.
Though modest by biotech standards, the investment represents a significant step for a pair of prominent Silicon Valley firms
that have yet to wade deeply into the life sciences. 1910 Genetics also won a $4.1 million seed round led by a personal investment from Y Combinator and OpenAI’s Sam Altman.
“We share the belief that life sciences is at an inflection point and that it’s a little bit behind on its uses of (machine learning) and automation,” Playground Global general partner Jory Bell told Endpoints.
Nwankwo presented a convincing case to lead the push to catch the field up, he said: “Jen is an extremely compelling founder. She had experience in shepherding drugs to market, business development in that context, and had put together an incredible team on the computational side.”
1910’s pitch is familiar to anyone who’s been following the field for the last decade: New artificial intelligence tools can shorten some of the most difficult and labor-intensive processes in drug development, including finding a good molecule to hit your target and then tweaking that molecule in potentially hundreds or thousands of different ways too.
They differ, though, in a couple of different ways. First, without delving into specifics, Nwankwo said the company’s approach for screening molecules doesn’t use the machine learning tool that has driven the AI revolution over the last decade and sits at the heart of AI drug discovery companies such as Atomwise and Exscientia.
Second, instead of helping with one major step of the process, Nwankwo and lead AI scientist Brandon Moore have developed a system of sequential and interlocking algorithms for different parts of the drug discovery process. SUEDE screens through 14 billion molecules that can hit a target. BAGEL uses neural networks to generate tweaks that make the molecule more drug-like. And CANDID uses neural networks to predict how it would perform on multiple metrics, such as solubility.
Like a growing number of AI companies, they also have a wet lab to quickly test out their best candidates. Nwankwo said they are also working on a platform, called ROSALYND, to apply AI techniques to protein-based drugs. They’re looking to find better ways of predicting a protein’s function from its sequence.
“Can AI help us there?” she said. “And if we’re going to fail, can we fail faster?”
Neither Nwankwo nor Playground’s Bell were shy about the company’s ambition. Although the financing is small and the company is only at 14 full-time employees, they’ll look to scale to 30 employees quickly. They already have programs in ophthalmology, infectious disease, neurology, immunology and aging.
Nwankwo said they have unnamed partnerships with large pharma companies. Eventually, they’ll have an internal and external pipeline, similar to a handful of other major computational companies.
“If you look at what Bruce Booth and the team did with Nimbus, it’s sort of similar from a business model perspective,” Nwankwo said. “We’re now coming to a point where we want to pick which areas we actually want to stick our neck out on and take those to IND. And you can expect us to come out this year and say Area A, Area B is where 1910 will focus.”
See the full article at EndPointsNews:
About 1910 Genetics: https://www.1910genetics.com/