CHICAGO (GenomeWeb) – A three-year-old project at Fred Hutchinson Cancer Research Center in Seattle has successfully applied a data visualization tool to all 32 data sets in the Cancer Genome Atlas.
Now, the developers are preparing to release a version suitable for clinical applications. Later this year, expect an update that produces 3D visualizations compatible with commercially available virtual reality headsets.
Called Oncoscape, this interactive, open-source platform is kind of an "experimental sandbox" for visualizing large sets of clinical and molecular data, according to its creator, Eric Holland, Hutchinson's director of human biology.
Hutchinson, led by Holland, started developing Oncoscape in 2014 and released version 1.0 to the open-source community about a year and a half ago. The most recent release, version 2.0, came about three months ago.
A 2016 paper in the Proceedings of the National Academy of Sciences showed that Oncoscape was successful in helping researchers discover tumor clusters based on gene sequencing, expression, and DNA methylation. "Our methods are generally applicable to all diseases and provide an intuitive means for physicians and bench scientists to work directly with 'big' biomedical data," Holland and two Hutchinson colleagues wrote.
Another article, published last month in the journal Acta Neuropathologica Communications, confirmed the efficacy of the technology in identifying diffuse gliomas as well as in predicting clinical outcomes for those with such tumors.
"We needed to come up with tools that would let us see all the molecular and clinical stuff together so we could manipulate it and understand it better," Holland said of the genesis of Oncoscape. "Research is obviously the first line, but eventually it needs to get built out and become operational" in clinical practice.
That is happening soon, now that Oncoscape has shown promise in the laboratory. But first, some work remains.
"In order for it to be clinically applicable, it has to be an easy mechanism for taking in someone's own data — an individual patient or collections of patients — and overlaying it," Holland said.
The data needs to come from both research collections and from electronic health records, according to Holland, a neurosurgeon who also serves as director of Seattle Translational Tumor Research at Hutchinson and of the Nancy and Buster Alvord Brain Tumor Center at the University of Washington Medical Center,
"You have to have a lot of input data like whole-exome sequencing and copy number," he explained.
At Hutchinson, all the data sets Holland has been working with have been derived from whole-genome or whole-exome sequencing, though that is changing. "We're coming up with shorter lists of genes that potentially still have the same kind of information in them so it will be easier to acquire that data cheaply," He said.
"If you have that data, regardless of whether it's CLIA-certified in the acquisition or not, we just need an easy way of overlaying that data. We have it, but it's not publicly available yet."
It should be released in the next month, Holland said, assuming he and his Hutchinson colleagues solve some thorny issues. "In order to have an overlay, you have to [address privacy issues] so no one else can see the data you're overlaying. There are a lot of little nuts and bolts that need to be set up in order to do this in a confidentiality-appropriate way," he said.
"It needs to be public data because this is a public site, and it needs to have the right consents around those patients because you can't just post their data, even deidentified, without making sure their consent was right. And then, of course, the owners of the data need to be OK with it being put out there," Holland continued.
On the heels of that update, expect to see version 3.0 of Oncoscape in a few months, featuring faster computation, 3D visualizations, and compatibility with consumer-grade VR systems. "That's going to be another quantum leap in the way you manage the data — better graphics, better ways of querying one big data set versus another — and much faster computation," Holland said.
"It's all being built in a way that's completely compatible with those [consumer VR] sets. It really makes you see all that data quite differently. You open the world up in VR and you can manipulate and look around and zoom in and out of the clusters," Holland said.
Another research paper is just about ready for submission, and Holland believes that almost all of his academic writing going forward will revolve around this visualization technology, particularly for gliomas and other brain cancers.
Meanwhile, the Oncoscape team is setting up the platform to support various applications from Hutchinson as well as from third parties. Five are up on the Oncoscape website now, and Holland demonstrated one called Markers and Patients at the Bio-IT World conference in Boston last month.
"I'm imagining them in a fantasy world of these things growing a little bit like iPhone apps, where different people will write their own application and add it to the collections. They would help build the infrastructure," Holland said.