Fighting cancer at 100 Gigabits per second

We often ascribe great, life-changing powers to high-speed Internet connections when it comes to how we communicate and consume content, but can they cure cancer? The newly formed Chan Soon-Shiong Institute for Advanced Health thinks so, and it’s investing hundreds of millions of dollars in a nationally distributed computing system to make it happen. At the core of its efforts is the National LambdaRail (NLR), a research network consisting of more than 12,000 miles of fiberoptic cable and capable of 100 Gigabits-per-second speed.
The CSS Institute, which launched earlier this year and is the brainchild of billionaire physician and biotech entrepreneur Patrick Soon-Shiong, wants to build a national health care infrastructure that will connect hospitals, practitioners and researchers to a network of high-performance gene-sequencing appliances dispersed throughout the country. Given the current disconnect between these stakeholders,¬†Bob Peirce, an advisor to Dr. Soon-Shiong (and former British diplomat), said it now takes an average of 17 years “to get from new science to new therapy.”
Peirce said the CSS Institute is focusing on cancer initially because cancer cells are different for everyone, which means a full gene sequence is often beneficial to accurately diagnose and treat individual patients. It also means cancer data is an ideal target for researchers using big data tools to identify trends and possibly discover broadly effective treatments. Ultimately, the institute envisions creating a self-learning health care system that’s constantly reassessing what it knows in order to give doctors and researchers up-to-date information.

Big data needs big networks

But gene sequencing is data-intensive — around 500GB per sequenced gene genome — which means a standard connection between facilities just won’t cut it. That’s why the CSS Institute actually took over financial responsibility for operating and upgrading NLR, an investment that should total about $100 million and that made Dr. Soon-Shiong the organization’s chairman and CEO. Much of that will go toward building out the genomics infrastructure, but NLR will still serve its existing function as part of the research networks for CERN’s Large Hadron Collider project and numerous U.S. research centers.
Cisco recently signed on as a partner (s csco) to help build out NLR’s 100 Gbps capabilities.

On the computing front, the institute has partnered with IO Data Centers, which both operates facilities and sells modular data center building blocks. IO’s Phoenix data center is the main hub of the CSS Institute’s infrastructure, where it houses its prototype sequencing appliance composed of hundreds of high-end multicore Intel (s intc) processors running on HP (s hpq) servers. It’s capable of running 50 gene-sequencing workloads a day, but NLR CTO Phillip LaJoie told me that, within a year, it will be part of a 10-node system dispersed in IO modular data centers across the country and capable of 500 sequences a day total.
As Stacey Higginbotham wrote recently, massive computing power and 100 Gbps connections are essential components of “the microscope of the 21st century” because they enable fast processing and easy sharing of petabytes of data that will lead to scientific advances similar to what the microscope enabled centuries ago by improving on the naked eye.

It’s not all about HPC

However, the end goal of any health care initiative is treating individual patients, a process that doesn’t always rely on supercomputers but that certainly can benefit from high-speed connectivity. Peirce said another component of the CSS Institute’s plan is to connect individuals’ health-monitoring devices to the computing system, which will help doctors monitor patients even when they’re at home and could help them catch warning signs before patients even get sick.
Additionally, Peirce said, the high-speed infrastructure will let doctors and researchers collaborate in new ways around high-grade imaging. Colleagues across the country could share and discuss 5-D images (which, like the one below, add the additional dimensions of time and movement) in real time without worrying about serious delays in rendering the images. Perhaps not coincidentally, NantWorks, a company owned by Soon-Shiong recently bought QImaging, a camera manufacturer that specializes in just such images. Ziosoft, a 4-D and 5-D imaging and analysis company. (NantWorks is renaming the company Qi Imaging after the acquisition.)
The CSS Institute’s mission is laudable, but one has to wonder how all-encompassing its national infrastructure actually will be. Already, Dell (s dell) has teamed up with several institutions to conduct genomics research for a particular type of childhood cancer, and there are even startup companies such as DNAnexus leveraging cloud computing to try to connect doctors and researchers with sequencing technologies and data. If we’re going to really put the power of computer science to improving medicine, it would seem that all the disparate projects need to figure out a way to work together, either under the same banner or at least by sharing data.
Feature image courtesy of Flickr user micahb37; 5-D heart image courtesy of the National Institute of Biomedical Imaging and Bioengineering.