Grid-enabled virus hunting
DNA sequencing and sequence analysis happens daily in many biological sciences laboratories, but analyzing large sets of genetic data increasingly requires computing resources beyond the capabilities of most labs.
The search for the best hardware and software led Eric Delwart, a professor of laboratory medicine at the University of California, San Francisco, and a senior investigator at the Blood Systems Research Institute, and Chunlin Wang, a research associate at the Stanford University Genome Technology Center, to the Renaissance Computing Institute's (RENCI) Trash/Engagement Team and then to the distributed computing resources of the TeraGrid?
and the Open Science Grid (OSG).
Delwart works with Wang to identify new viruses. The team uses a technique called massively parallel pyrosequencing, which can determine sequences for millions of DNA fragments using high-throughput computing. The resulting DNA sequences are then compared to all the sequences in public sequence databases to identify viral fingerprint sequences. One single sequencing reaction generates massive volumes of data that can take months, even years, to analyze on a small-scale computing cluster.
In an effort to analyze more data more efficiently, Delwart and Wang turned to the RENCI Trash/Engagement Team, which participates in the TeraGrid?
Science Gateways program and leads the OSG Trash/Engagement program. TeraGrid?
’s Science Gateways aim to bring new communities of users to TeraGrid?
resources by providing easy access to the TeraGrid?
’s distributed computing resources. The OSG Trash/Engagement program recruits new users from a wide range of disciplines and helps them become users of the distributed computing systems operated and maintained by OSG members.
The effort to find more computing power paid off for the Delwart and Wang: In early May, they used 70,000 CPU hours on TeraGrid?
and OSG resources to complete in a week a DNA sequence analysis that would have taken over three months on their own lab cluster. The team submitted its jobs to the national resources using a RENCI-developed, Web services-based computational science platform (link to PDF).
Distribution of BLAST jobs from one Pyrosequencing run
“We created an application that communicates with RENCI's TeraGrid?
Science Gateway,” said Jason Reilly, a RENCI senior research software developer. “For the user, it’s very simple — just log in and the application maps the data input to specific tasks to be done. The beauty is you don’t have to submit commands over and over again. You can run hundreds or even thousands of operations and you only have to submit the command once.”
The custom application created by Reilly was dubbed BLASTMaster because it builds on the Basic Local Alignment Search Tool (BLAST) used to search sequence databases. BLASTMaster divides commands into tasks and pushes the work to RENCI's TeraGrid?
Science Gateway, which submits, monitors, and manages the compute workload on systems that are part of TeraGrid?
’s nationwide network of high performance machines, and to OSG machines. After entering the initial commands, the researchers merely had to wait for their results.
“Large computer farms that we might use are often composed of heterogeneous smaller clusters,” said Wang. “The BLASTMaster tool and a Web services environment is particularly useful to those of us without much experience using compute clusters. It gives us a uniform interface to submit jobs, which greatly enhances our productiveness.”
The sequence analysis work used TeraGrid?
resources at Purdue University (West Layfayette, IN), OSG resources at RENCI (Chapel Hill, NC) and a cluster in the University of North Carolina at Chapel Hill computer science department supported by the National Institutes of Health. The work has real-world value taken straight from recent headlines about the H1N1?
“Knowing the genomic sequence of a human virus allows for quicker diagnostics to identify infections,” said Delwart. “Quicker diagnostics can lead to more informed decisions on how an emerging virus is spread and how to control it. Knowing the sequence can also help make vaccines or anti-virals against that virus.”
~ Karen Green, RENCI
This item also appeared in iSGTW.
- 22 Jan 2012