Artificial intelligence is coming to genetic sequencing.
Gene-sequencing company Illumina is teaming up with IBM Watson Health to put Watson’s artificial intelligence to work analyzing Illumina’s solid tumor sequencing panel.
Here’s how it works: A cancer doctor will take a sample of a person’s solid tumor, which will then be sequenced using Illumina’s “TruSight Tumor 170” panel, which will screen for 170 different tumor mutations. Once that happens, Watson will get to work analyzing them and giving back a report to the oncologist, giving them information about the mutations the test found and what the next step might be.
“We’re partnering with IBM Watson to take data that comes out of those tests and leverage deep learning algorithms and and cognitive learning algorithms to be within minutes be able to provide an actionable report for the oncologist,” Illumina CEO Francis deSouza told Business Insider. The hope is to open up what deSouza referred to as a “bottleneck” that’s kept more people from getting their tumors sequenced.
“The intent of this relationship is to address the vast majority of the market, where cancer patients are treated,” de Souza said. That means being used in community hospitals, rather than cancer institutions or research hospitals.
The Watson software is expected to be ready to go for the panel in early 2017, the companies said in a release.
Other companies, such as Foundation Medicine, deliver these kind of reports based on their own genetics panels, and Watson Health has also teamed up with Quest Diagnostics for its own Watson-powered tumor sequencing analysis.
Machine learning in genomics
This isn’t to say machine learning isn’t being used already by Illumina. DeSouza said Illumina already uses big data techniques such as AI to process the sequencing information that comes into Illumina’s cloud, and as a way to keep tabs on how its instruments are doing out in the field.
“This problem required a whole different scale of capabilities.” deSouza told Business Insider. That’s based in part on the data that Watson was trained on.
Guardant Health, a company that makes blood tests that screen for tumor mutations, often called “liquid biopsies,” is another company that’s using machine learning to make its test more sensitive to mutations in the blood. Guardant CEO Helmy Eltoukhy told Business Insider that one of the problems is getting enough data in place to get more accurate in the reports.
“It’s like we’re populating the Internet while at the same time creating Google Search,” he said.