Researchers are always looking for a better way to turn the flood of genomic data spit out by their DNA sequencers into better treatment options for patients. The New York Genome Center is getting the first crack at a potential answer to that problem: Watson.
Today, the NYGC, the big genomics hub formed by 12 of the New York’s research institutions last year, is announcing a partnership with IBM. Through the deal, the NYGC will be able to use a custom-made prototype version of IBM’s Watson as a genomics research tool—the first time the tech giant’s cognitive supercomputer would be used in such fashion.
By teaming the NYGC’s sequencing capabilities with Watson, the idea is to not only speed up the process of turning genomic insights into personalized treatment regimens, but also that the IBM computer will be able to integrate new data and turn up potential cancer treatment options that researchers would’ve otherwise had a hard time finding, according to Toby Bloom, the NYGC’s deputy scientific director of informatics.
The deal is essentially a test run. The NYGC will use Watson to try to help oncologists tailor specific drug regimens to individual patients with glioblastoma—an aggressive form of brain cancer that typically kills people 12 to 14 months after they’re diagnosed—based on the specific genetic mutations in each patient’s tumor. The goal is to use this process to improve the standard of care for glioblastoma patients, and then expand the NYGC/Watson effort into other types of cancer. Bloom says nine hospitals and 20 neurologists, all in New York, are on board so far, but that the NYGC isn’t limiting the effort to any particular hospitals.
Here’s how the process is expected to work. Once a patient is diagnosed with glioblastoma, the NYGC will get a tumor sample. Researchers there will sequence the tumor’s genome, and analyze it to come up with a set of mutations that look like the “drivers” for that specific patient’s tumor, Bloom says. Researchers will then feed that list of mutations into Watson, which will do all of the computational dirty work to find a list of drugs that would target them specifically.
Watson’s job will be “to do the literature search, the drug database search, and find all the relationships between those specific mutations and drugs that are available or maybe in clinical trials that they can find, or even drugs that are related to the pathways that those mutations are associated with,” Bloom says.
From there, a team of neurologists and pharmacologists will look over the list of drugs that Watson generates, pick a set of them that they think would best work together, and recommend those to the patient’s doctor. Should the patient decide to follow the recommended regimen, the NYGC will track his condition, the rate at which the tumors spread, etc.—and then start feeding that information back into Watson.
“So we’ll be giving Watson back information about combinations of mutations, combinations of drugs, and the outcomes of using those drugs on those mutations,” Bloom says. “That’s a very complex set of associations, and we’re hoping that Watson, and Watson’s learning model, can then find information in that that we can not see so fast.”
The study hasn’t started as of yet, but the hope is that as more patients start undergoing treatment, and more data is fed to Watson, the computer will find things researchers wouldn’t, and “we can iterate more quickly and improve care more quickly,” Bloom says. Assuming the lists Watson produces leads to treatment regimens that helps patients keep tumors in check longer, and live longer—which has yet to be seen—the project would then expand. Presumably, Watson would also be used by other genome centers as well (the NYGC does not have an exclusive license).
“There are 1,700 glioblastoma patients in NY every year,” Bloom says. “So there’s a lot to be done in New York.”