 To achieve translational medicine—often defined as ‘bench-to-bedside’ research—healthcare organizations must evaluate their existing infrastructure boldly and assess the technology required to support exploding data volumes and compute-intensive applications and processes.
“To realize the vision of translational medicine, companies need very specific capabilities,” says Dave Medina, Director, HP Worldwide Pharmaceutical and Life Science Markets. “They need to collaborate across the enterprise, break down silos for improved information access and governance, access multiple data sources, and integrate genomic information into clinical records.”
Striving for translational cancer care
M.D. Anderson Cancer Center is dedicated to eliminating cancer with integrated programs in cancer treatment, clinical trials, education programs and cancer prevention.
“At M.D. Anderson, we start with the science, but ultimately, we focus on how to apply the results of the scientific work to the clinical workflow to improve patient care,” says Dr. Lynn Vogel (pictured, right), Vice-President and Chief Information Officer, M.D. Anderson. “For example, we are applying our understanding of genetic expressions to a patient’s propensity to respond to specific medications, which creates a direct link between the science and personalized treatment.”
When Krishna Sankhavaram (pictured, left), Director for Research Information Systems and Technology Development, joined M.D. Anderson in 2005, his department was brand new. The goal was to create an IT department to support and develop infrastructure and offer tools, applications and services for basic and translational sciences research. At that time, research had its own grants and funding and acted independently of IS in terms of technology.
“As I became more acquainted with the workings of the business, I realized that one of the resources we required was a high powered, high performance computing environment,” says Sankhavaram. “Most people were paying money to buy cluster platform cycles in other places and then cobbling them together as a cluster. There was no centralization or coordination with central IS; everything was distributed and ad hoc as research needs arose, which created duplications and challenges in collaboration and information sharing.”
High Performance Computing
Sankhavaram in cooperation with a faculty committee realized the need for a combination of memory and cluster technology as well as the centralization of storage and data archives. Sankhavaram also wanted a partner that had the life science expertise to optimize code and assist with scientific application development. After a detailed evaluation of various vendors, the committee chose HP to implement various hardware and software solutions.
M.D. Anderson went live with the HP High Performance Computing (HPC) solution in November 2006, and by February, Sankhavaram was running the system for 10 to 15 days at a time, at full capacity. A second cluster, double the size of the first install, went live in September 2007 to accommodate the exploding usage across the organization.
“The results were phenomenal,” notes Sankhavaram. “Between November 2006 and summer 2007, approximately 24 published articles came out from using our cluster as the primary computational resource. Prior to that, we published one or two papers in a year.”
Within the HPC environment, analyzing an experimental diagnostic imaging dataset went from 20 minutes to 20 seconds. Similarly, radiation physics took 72 hours to run a one billion particle dataset on its own cluster, which became a mere five hours using the HPC cluster.”
One more step toward translational medicine
M.D. Anderson extended its partnership with HP to develop an application platform to improve cancer care by linking molecular data with clinical information. Researchers will then be able to ask intelligent questions, discover, and assess how certain molecular markers behave in specific kinds of cancers.
|
Printable version
