A Seattle startup hopes AI will be a Rosetta Stone for cancer immunotherapy research
- Mike Pearl
Ozette sounds a little like Theranos, except plausible.
On July 28, a Seattle-based AI firm called Ozette, which focuses on life sciences research, especially on immunotherapies for cancer, announced a $26 million round of Series A funding. Ozette’s website describes its mission as “digitizing biology using novel computational methods and machine learning to give a full view of an individual’s immune system.” The company was spun off from a program at Seattle’s Fred Hutchinson Cancer Center.
The CEO of Ozette, Ali Ansary, remains an attending physician at the Fred Hutchinson Cancer Center, and took part in the Allen Institute’s “Entrepreneur in Residence” program. (The Allen Institute is the principal backer of pnw.ai.) His pitch for Ozette, delivered to Endpoints News, sounds a bit like the promise of the debunked biotech startup Theranos, except real: “You cannot go and always do biopsies on the tumor. In that case, you can extract valuable information around what the immune system is doing in that tumor, by just looking at the blood,” Ansary told Endpoints.
The lead funder in this $26 million round was Madrona Ventures, also based in Seattle, and the firm itself was incubated at The Allen Institute for AI.
The technology that the Ozette website describes as “digitizing biology” concerns cytometry, a method used in the field “to interrogate the state of an individual’s immune system at the single-cell level,” according to a 2021 study of a similar technology by Fred Hutchinson researchers. That study demonstrates the viability of Ansary’s purported blood analysis method as an alternative to a biopsy. Its lead author, Evan Greene, is now vice president of quantitative methods and analysis at Ozette.
What Ozette envisions is an AI-based analysis system for identifying the cells present in a tissue sample — say, during a trial for a promising immunotherapy that might be useful in treating cancer — performed by quantifying the 30 to 40 key proteins involved in cytometry.
According to the Endpoints story, the way this is presently done is that a sample is analyzed manually by a computational biologist. The computational biologist’s data goes to a data scientist. The data scientist’s analysis goes to a statistician, and an educated guess can be made as to what cells are present.
Ozette then hopes to produce a machine learning-based software platform that digitized samples can be fed into, and all that analysis, which normally takes months, can be done in days. And, worth noting: if this company were run by convicted Theranos fraudster Elizabeth Holmes, Ansary and his team would be claiming this analysis would happen instantaneously. So Ozette has plausibility on its side.
Not only will researcher’s be able to, hopefully, use Ozette’s technology to perform this type of analysis more quickly, but, Ansary told GeekWire, he hopes to expand the possibilities for research perhaps exponentially. Ozette may have the potential to solve a problem in which different researchers, performing similar experiments involving different hardware are, at present, unable to share data.
If Ozette can be a sort of Rosetta Stone for this type of research, “it’s here where the magic happens,” he said.