Synthesize Bio, a Seattle-based startup founded by leaders from Fred Hutchinson Cancer Center, announced $10 million in funding from Madrona Venture Group.
The biotech company aims to make new drug discovery faster and cheaper by using artificial intelligence to simulate the results from hypothetical lab experiments. The team built what it’s calling a generative genomics foundation model (GEM-1) to predict gene expression, providing insights into how a novel drug is expected to impact cell behavior.
On Tuesday, Synthesize Bio researchers published a preprint on bioRxiv explaining the model and its performance, which the scientists said “predicts future gene expression experimental results with lab-level accuracy.”
Synthesize Bio has a 16-person team and was co-founded nearly two years ago by Fred Hutch Chief Data Officer Jeff Leek and Robert Bradley, director of the Translational Data Science Integrated Research Center at Fred Hutch.
The team trained its model on publicly available information correlating research experiments with RNA sequencing data — which captures the genes being expressed in a cell. The dataset included healthy and diseased human cells and tissue that was analyzed in lab studies and clinical trials.
“This is the kind of infrastructure shift that redefines what’s possible in life science R&D,” Matt McIlwain, managing director at Madrona, said on LinkedIn. “With generative genomics, researchers can ask bolder questions, design smarter trials, and test hypotheses that were previously unaffordable or impossible to explore.”
In its preprint article, the Synthesize Bio team elaborated on the problem it is trying to solve. “Many lab experiments can be conducted no faster than the speed with which cells grow or diseases develop,” the company wrote, “while clinical trials are similarly governed by the availability and recruitment of participants and essential regulations to protect patients.”
The generative capabilities of AI models, they continued, “offer the possibility of circumventing these fundamental biological constraints by computationally simulating many, or even all, limiting experimental steps.”
The technology can also provide incremental feedback during the research process, guiding scientists to more promising solutions.
Synthesize Bio, for example, could help researchers whose drug candidates have cleared clinical safety testing assess whether those therapies are also likely to deliver meaningful health benefits.
“Biologists, whether they’re doing fundamental science or they’re working at a company trying to advance a drug through clinical trials, they’re constantly in a position where you have to make critical decisions,” Bradley told GeekWire, “and frequently they’re very costly decisions without enough information.”
The technology, which is now available for anyone to use, comes at a challenging time as the Trump administration is working to slash grants and other funding for research as it aims to cut government spending. Leek and Bradley emphasized that support for basic laboratory research remains essential.
“Our vision is not that somebody will start using our model and do the exact same work that they’re doing right now, but for less money,” Bradley said. “Our vision is that somebody will start using our AI model and be able to do 10 times the science and make 10 times the discoveries that they are right now.”
McIlwain praised the team for its cautious approach and rigorous testing of the technology.
“These two are healthy skeptics,” he told GeekWire. “They’re pragmatic, grounded folks that are like, ‘prove it to me, show it to me.'”
Synthesize Bio joins a flurry of biotech startups and research groups developing AI-powered systems in the Seattle area, including:
- The University of Washington’s Institute for Protein Design (IPD) uses AI to build novel proteins that could be used in treating wide-ranging diseases.
- The Fred Hutch-led Cancer AI Alliance, a consortium that also includes Dana-Farber, Memorial Sloan Kettering, and Johns Hopkins with support from the Allen Institute for AI (Ai2) and Google Cloud.
- The Allen Institute’s Seattle Hub for Synthetic Biology, which is using a DNA-based technology to make a recording of what a cell experiences over time.
- Startups including Xaira Therapeutics, Archon Biosciences, Lila Biologics, Outpace Bio, A-Alpha Bio, Talus Biosciences, Potato and many others. IPD alone has spun off 10 startups.
Other authors of the preprint titled “Generative genomics accurately predicts future experimental results” are: Gregory Koytiger, Alice M. Walsh, Vaishali Marar, Kayla A. Johnson, Max Highsmith, Alexander R. Abbas, Andrew Stirn, Ariel R. Brumbaugh, Alex David, Darren Hui, Jeffrey M. Kahn, Sheng-Yong Niu, Liza J. Ray, Candace Savonen and Stein Setvik.
Editor’s note: Story updated Sept. 17 to include comments from Synthesize Bio’s co-founders and investor Matt McIlwain.
