The Week in Review
Here is a collection of highlights from this week’s news stream as reported by HPCwire.
Supramap tracks infectious diseases, predicts outbreaks
This week researchers from the Ohio Supercomputer Center and the American Museum of Natural History revealed an online tool, called Supramap, that tracks infectious diseases, such as the avian influenza virus (H5N1) and the H1N1 virus.
Supramap uses parallel programming on high performance computing systems at the Ohio Supercomputer Center (OSC) to track the diseases across time, geography, and host animals, including humans. The genetic data is gathered and displayed in online geographic information systems, such as Google Earth. The application has been described as a “weather map for diseases” and will be a boon to public health professionals in helping predict disease outbreaks.
The research was published in the April 2010 online issue of Cladistics. Daniel A. Janies, an associate professor at Ohio State University, is the first author of the paper.
Supramap does more than put points on a map — it is tracking a pathogen’s evolution. We package the tools in an easy-to-use web-based application so that you don’t need a Ph.D. in evolutionary biology and computer science to understand the trajectory and transmission of a disease.
We have come to the point in computational genomics where we have generated a lot of raw data. Lately, we are seeing more and more applications being developed to help us analyze that data in practical, meaningful ways, putting it to work to benefit humanity. Supramap is just such a tool.
Genetic applications move forward
Continuing the progress on the genetic front, there were several other stories of interest announced this week.
Researchers at the Virginia Bioinformatics Institute have located small genes that were missing from the scientific knowledge base. Using grid computing, and the mpiBLAST computational tool, researchers uncovered the information in only 12 hours, instead of the 90 years that the same task would have taken using personal computing resources.
This is the first large-scale attempt to find missing information in the GenBank DNA sequence repository. The study was reported in the journal BMC Bioinformatics.
An announcement from Ohio State University calls attention to a computing process that helps identify molecular structures that have the most promise to serve as the active component in new medications. Finding these molecules is not easy, but researchers are using computer simulations that connect a part of the molecule, called fragments, with a part of the diseased protein, called “hot spots.”
In order to get the fragment candidates, massive computing power is called into play to narrow down the huge pool that is gathered from thousands of existing drugs already on the market. Two computers are tasked with the simulations: one in Ohio State’s College of Pharmacy and the other in the Ohio Supercomputer Center.
Canadian researchers are using the resources of the High Performance Computing Virtual Laboratory (HPCVL) to scan the entire Homo sapiens proteome. This will be the first time such a task has been completed. From the announcement:
The proteome is the complete set of proteins produced by a species. Protein-protein interactions are an integral part of many biological processes within the body’s cells, including signaling processes to respond to outside stimuli such as the level of oxygen in the environment, transporting nutrients, and responding to threats from viruses such as H1N1 and HIV. Knowing the interactions in cells — at the molecular level — is essential for understanding cell behavior. It is also essential for understanding the impact of pathogens on cells.
The researchers expect to have complete results in a few months.
The last story of the bunch concerns the promising frontier of personalized medicine. A paper was published in Nature this week by over 200 members of the International Cancer Genome Consortium (ICGC), detailing a huge leap forward in cancer research. The group has been cataloging the genetic changes of the 50 most common cancers, with 500 genomes from each cancer type, and will now make that information publicly available online.
This research has the ability to revolutionize cancer treatment by using the genetic makeup of a patient’s specific cancer and pairing that with a treatment that has been proven to work on a similar cancer. I say similar because no two cancers are the same, but they share similarities. This will help patients get the most effective treatment first, instead of having to go through many trials before reaching an effective medicine or procedure. The information will also assist with enacting more effective and more timely clinical trials.
Said Professor Andrew Biankin, member of the Nature paper’s writing team, and part of the Australian project arm of the ICGC:
“The consortium’s Internet-based databanks will help us treat specific cancers with specific treatments. Not only that, the information will help us understand why some treatments work and others do not, and then design better drugs to target faulty elements or mechanisms.”
The Internet databanks will have different tiers of access to regulate who can see what. The general public will be able to view summaries of the data, whereas medical and research professionals will be able to request detailed accounts.
Second annual DICE Awards program shines light on Tesla GPU and Spectra Logic
The winners of DICE’s second annual Data Intensive Impact Awards were announced today. DICE stands for Data Intensive Computing Environment, which is the moniker for Avetec’s High Performance Computing Research Division.
The Future Technology Category award went to NVIDIA’s Tesla 20-series GPU, which the DICE team considers a critical new technology for the HPC space:
When compared to the latest quad-core CPUs, Tesla 20-series GPU computing processors deliver equivalent performance at 1/20 the power consumption and 1/10 the cost. More importantly, Tesla GPUs enable high performance computing users to scale their computing resources to get significant boosts in performance while staying within tight power and monetary budgets.
And the Product Category award went to Spectra Logic for its T-Finity enterprise tape library, which scales to more than 45 petabytes in a single library and to more than 180 petabytes in a single, unified library complex.
Said Al Stutz, Avetec CIO and DICE team leader:
“Our team selected the T-Finity for its versatility in data intensive environments. The product helps with the demanding archiving and backup environments experienced in the enterprise IT, federal, high performance computing (HPC) and media and entertainment space.”
Last year’s awards went to BlueArc and the now-defunct Woven Systems.
Avetec is a not-for-profit research company that aims to advance the competiveness of American companies. DICE conducts technology testing and validation for new and emerging HPC data management solutions, and the Data Intensive Impact Awards highlight products and technologies that have enabled progress in HPC data management in locality, movement, manipulation and integrity, as well as power and cooling efficiencies.