May 21, 2018
SAN FRANCISCO (CN) – Bill Paseman has two choices for treating his rare and deadly kidney cancer: do nothing or let 200 scientists from around the world analyze his DNA to uncover clues for promising new treatments.
That’s because there are no effective treatments for his late-stage papillary renal-cell carcinoma type 1. Little is known about the genetic drivers of papillary renal-cell carcinomas, which account for just 15 to 20 percent of adult kidney cancers. And because the patient market is small, pharmaceutical companies don’t focus on the condition.
So this past weekend, 150 computational biologists, geneticists, oncologists, artificial intelligence researchers, and computer developers from top universities gathered to analyze the genomes of Paseman’s kidney tumor and his blood – with Paseman present. A team from Harvard Medical School joined in the effort online, as did researchers from Canada, India and Mexico.
The immediate goal of the “hackathon” was to identify mutations in Paseman’s DNA fueling his cancer, and to pinpoint drugs developed for other cancers that might help him should his disease progress. Paseman expressed enthusiasm over one team’s discovery of five different drugs for his condition already approved by the U.S. Food and Drug Administration. The team identified the drugs using a machine learning method called deep learning to analyze one of Paseman’s mutations.
By comparing a patient’s rare cancer with more common ones that are better understood, researchers can ultimately develop new drugs for diseases like Paseman’s.
“My goal is not to get the Nobel prize for this,” said Paseman, 63, a tech entrepreneur who cofounded Rarekidneycancer.org after he was diagnosed in February 2014. “My goal is for one of the hackathon attendees to get the Nobel prize and for me to be in the audience, applauding, while I’m taking the tablet for my condition.”
This was the second hackathon on a rare disease run by the National Institutes of Health and organized by San Francisco nonprofit Silicon Valley Artificial Intelligence, a group that aims to accelerate disease research through collaborations between biomedical and artificial intelligence experts. Last year, hackathon participants analyzed DNA from a patient’s neurofibromatosis type 2 tumor, or NF2. The condition produces benign brain tumors that steal a patient’s sight, hearing and ability to walk.
The hackathon format can help fill the gap in rare-disease research and speed the development of new drugs. Jyotika Varshny, one of last year’s winners, incorporated her results into her postdoctoral research at University of California, San Francisco Medical Center and started a company that develops biosimulation models for testing experimental drugs, eliminating the need for animal testing.
Onno Faber, the patient whose DNA was studied that year, also launched a company that pools and analyzes patient data for NF2 to identify genes driving the disease.
“It’s fascinating putting data out there, expanding the ability of people to collaborate,” said Peter Kane, chief executive of Silicon Valley Artificial Intelligence. “It far exceeds what a single company or lab at a university could potentially do that is still constrained by a group of 5 to 15 people.”
However, it’s difficult to develop new treatments for rare diseases because there isn’t enough genetic data available for researchers to study. According to Varshny, numerous NF2 clinical trials have failed because researchers lack the data required to perform complex statistical analyses.
To get around that, Varshny’s team used genetic data from an NF1 tumor to identify potential causes of NF2. Although the diseases are similar, more data is available for NF1.
He also wants patients to donate their tumor tissue for DNA sequencing and research. But most genetic data is sequestered in private databases around the world and not shared, according to David Haussler, a prominent geneticist at the University of California, Santa Cruz.
“The generalization is lacking to really approach that,” Varshny told the crowd on Friday night. “We have to build something more scalable.”
Paseman envisions using additional hackathons to build a repository of research on rare cancers. All of the data, code and methods developed for a particular hackathon patient can be used to study future hackathon patients, and vice versa.
“This is an enormous problem and an enormous lost opportunity,” Haussler said in a 2015 TED talk at the university.
[ Read More ]