In an unprecedented genetic research effort, over half a million U.S. veterans' genetic makeup has been mapped out in a study crucial for the future of precision medicine. This comprehensive project, harnessing one of the most advanced supercomputers in the world at the time of its operation, was recently spotlighted for its strides in understanding the genetic underpinnings of countless health conditions. According to a statement from Argonne National Laboratory's senior computational scientist and study's senior author Ravi Madduri, "It’s not only one of the largest genome-wide association studies ever done, but it’s analyzing some of the most diverse data ever assembled."
The major undertaking, known as the Department of Veterans Affairs' Million Veteran Program (MVP), amassed genetic data from 635,969 veterans of diverse ages, races, and backgrounds. Tasked with analyzing this massive pool of genetic data, scientists turned to the Department of Energy’s Oak Ridge National Laboratory (ORNL) and its Summit supercomputer for computational support. Working with the data, which was originally prepared for a different kind of computer, required converting it to be optimized for the GPUs that powered Summit, "a task that took years on its own," affirmed Madduri in the course of explaining the challenges.
Summit's capabilities allowed for 200 quadrillion calculations per second, which made it possible to accurately associate the genetic data with more than 2,000 health traits, without sacrificing detail. "The diversity and scale make this study stand out, and we couldn’t have done this study with any other set of computers," said Madduri, highlighting the unique capacity of Summit to deal with such vast genetic information.
Madduri told Oak Ridge National Laboratory that the research calculated associations between nearly 44 million genetic variants and more than 2,000 traits, culminating in a total of 26,049 associations between genetic variations and health traits. In a sea of data comprising more than 30 terabytes, the raw information held the equivalent volume of around 200 million pages of text. Even more impressively, this mammoth analysis ran for over 500,000 node-hours, leading to more than 350 billion examinations of associations.
The results of this unparalleled study will, according to ORNL officials, be made accessible for future research via the National Institute of Health’s National Library of Medicine. ORNL has also developed a computational pipeline and an online platform called CIPHER to visualize the data for other researchers – all provided through a collaboration with the VA and supported by the Department of Energy's Office of Science’s Advanced Scientific Computing Research program. Offering a testament to the project's success and potential impact, Anuj Kapadia, overseeing advanced computing for health sciences at ORNL, expressed pride in advancing the well-being of our veterans and driving "groundbreaking discoveries with far-reaching impact on human health."
