New supercomputer can analyze huge volumes of genomic data
Supercomputer Hyperion and Jason de Koning, PhD, computational and evolutionary geneticist in Faculty of Medicine. Photo by Bruce Perrault
A new supercomputer at the university is helping genetics researchers unlock the secrets of evolution and the genes that cause disease. The work is being done by Jason de Koning, a computational and evolutionary geneticist in the Faculty of Medicine and a member of the Alberta Children’s Hospital Research Institute (ACHRI).
De Koning has just installed the computer known as Hyperion, named for the mythological Greek god of light. “We’re very excited about the potential for this technology to enable sophisticated analyses of huge volumes of genomic data,” says de Koning, who moved from the University of Colorado in 2012 as a recruit by ACHRI to establish a research group in bioinformatics and computational genomics.
Algorithms put to test in supercomputer
The computer (cluster) is capable of 30 trillion calculations per second. Hyperion’s capabilities come from its 10 Xeon Phi accelerators, the newest technology on the market. Each accelerator has 60 enhanced computing processing units on one chip. The funding for the technology was provided by a $116,000 grant from the Canada Foundation for Innovation (CFI) and matching funds provided by ACHRI and Dell.
The computer is now performing its first computations, providing the infrastructure for de Koning’s research into the genomic variations in humans which lead to disease susceptibility. To find these variations, de Koning develops new algorithms. “Technological advances have enabled enormous amounts of genomic data to be sequenced worldwide. These data contain rich information on the effect of genetic mutations on biological functions, the origins and history of the human species and on the genetic causes of disease,” says de Koning.
“We’re interested in the computational problems that arise when extracting this information from large volumes of data. Bigger computers are part of the solution, but scalable algorithms are actually far more important. Hyperion is facilitating our work into new algorithms for large-scale inference that bridge the ‘Big Data’ emerging in genomics with detailed biological models.”
As well as performing this computational research, de Koning is an evolutionary geneticist. Last year, he published his most recent research in the Proceedings of the National Academy of Sciences. He and an international team successfully sequenced and characterized the entire genome of the Burmese python – a complex vertebrate with extreme adaptability.
“The sequencing of new species with unusual adaptations helps us to understand how new functions evolve and how genetic changes lead to phenotypic changes, which are two of the most fundamental problems of modern biology.”
But the utility of comparative genomics is not limited to improving an understanding of basic biology. Everything in genomics is about making clever use of comparisons – such as between healthy and sick individuals, or between humans and other species. A single genome by itself says little. But many genomes together can say a lot.
“One of the most important frontiers in genomics is the development of new ways to better exploit knowledge about evolution,” says de Koning,whose lab is currently working on several projects that aim at developing these new computational approaches.
In 2012, the Alberta Children’s Hospital Research Institute opened a genomics and bioinformatics facility with a $5.5 million donation from the Alberta Children’s Hospital Foundation and a $2 million Faculty of Medicine contribution.
ACHRI’s Next Generation Sequencing and Bioinformatics Platform is aligned with the University of Calgary’s Eyes High research platform Analytics and Visualization. This platform focuses on being able to handle vast amounts of data fast and effectively allowing researchers and faculty to work with large sets of data to better capture, analyze and visualize information.