November 23, 2015
The followin article is from the NIH Director's Blog (directorsblog.nih.gov.):
After graduating college with degrees in physics and computer science, Amanda Randles landed her dream first job. She joined IBM in 2005 to work on its Blue Gene Project, which had just unveiled the world’s fastest supercomputer. So fast, in fact, it’s said that a scientist with a calculator would have to work nonstop for 177,000 years to perform the operations that Blue Gene could complete in one second. As a member of the applications team, Randles was charged with writing new code to make the next model run even faster.
Randles left IBM in 2009 for graduate school, with the goal to apply her supercomputing expertise to biomedical research. She spent the next several years developing the necessary algorithms to produce a high-resolution 3D model of the human cardiovascular system, complete with realistic blood flow. Now, an assistant professor at Duke University, Durham, NC, and a 2014 NIH Director’s Early Independence awardee, Randles will build on her earlier work to attempt something even more challenging: simulating the movement of cancer cells through the circulation to predict where a tumor is most likely to spread. Randles hopes all of her late nights writing code will one day lead to software that helps doctors stage cancer more precisely and gives patients accurate personalized computer simulations that put an earlier, potentially life-saving bullseye on secondary tumors.
Ten years ago, Randles’ vision would have been impossible. But progress in understanding basic human biology together with advances in imaging technology, including high-resolution CT and MRI scans, now enable high performance computer scientists like Randles to design incredibly detailed simulations according to the precise geometries of a person’s circulatory system.
Supercomputing has also entered a new dimension of speed, and tomorrow’s machines will leave them in the dust. President Obama signed an executive order in July to launch the National Strategic Computing Initiative, which has the goal of pushing supercomputing to the exascale, or 30 times faster than today’s fastest machines, by 2025. The NIH is part of the team that aims to make this a reality.
Randles continues to make great strides in the laboratory. In a recent paper, her group reports it has mapped every blood vessel in the body that is 1 millimeter in diameter or larger, and improved its investigative resolution from 10 microns to 9 microns [1]. That’s very close to distinguishing individual red blood cells flowing through the vessels.
Her group has plans to model the fluid dynamics within blood vessels and simulate the movement of each individual red blood cell, white blood cell, and ultimately cancer cell. That will require accurately modeling the real-life contortions of millions of cells interacting under the shear stress conditions of a blood vessel. Getting it right will be critical in learning to simulate how long cancer cells will spend at particular points along the blood vessel walls and to predict the sites where they will ultimately lodge and produce secondary tumors.
It also will require patience. For full body simulations of the vascular system and the trillions of cells coursing through our blood vessels, Randles will need to draw upon an enormous amount of computational power that doesn’t yet exist—and likely won’t for the next few generations of supercomputers. It’s not too early to begin working toward this goal, and the outstanding work of Randles and her colleagues is earning praise. Their latest simulation of the human circulatory system, which I referenced above, was among five finalists for the just-awarded 2015 Gordon Bell Prize in High Performance Computing. Here’s wishing Randles and all of the finalists good luck in their research going forward; we are counting on them for continued innovation and progress.
Reference:
[1] Massively parallel models of the human circulatory system. Randles A, Draeger EW, Oppelstrup T, Krauss L, Gunnels JA.ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis. 2015 November 15-20.
Links:
Amanda Randles (Duke University, Durham, NC)
Metastatic Cancer (National Cancer Institute/NIH)
Gordon Bell Prize in High Performance Computing (Association for Computer Machinery, New York)
Executive Order-Creating a National Strategic Computing Initiative. White House. 2015 29 July.
NIH Director’s Early Independence Award Program (Common Fund/NIH)
NIH Support: Common Fund
This article originally appeared on directorsblog.nih.gov.