MIDDLEWARE
Details of the New Brain Imaging Research Network, Part I
By Steve Fisher, Editor In Chief -- Last week the National Center for Research Resources (NCRR) awarded more than $20 million to a consortium of universities coordinated by the University of California, San Diego (UCSD) to build the first nationwide high-performance computer environment to study diseases of the brain. Dubbed BIRN (Biomedical Informatics Research Network) a great deal of work will be done at the San Diego Supercomputer Center. In part one of a two-part interview, Supercomputing Online spoke with SDSC Director Fran Berman. Part two will feature the comments of BIRN project lead Mark Ellisman.
Supercomputing: Please tell us about the Biomedical Informatics Research Network (BIRN) project and how it ties in with NPACI's neuroscience & DICE thrust areas. BERMAN: The BIRN activity represents an important direction for NPACI and in particular the Neuroscience, Data-Intensive Computing Environments, and Interaction Environments thrust areas. One of the major challenges in BIRN research is data integration. Brain data sets from different labs, collected with different equipment, require incredibly sophisticated software and algorithms for permitting researchers to make comparisons among several brains. The techniques for doing so were first prototyped as part of a Neuroscience-DICE project on federating brain data. In addition, the size of the data sets presents a visualization challenge, and the scalable visualization toolkits being developed by the Interaction Environments thrust will form common infrastructure for the BIRN participants. BIRN also represents a practical application of NPACI's infrastructure building efforts. NPACI Rocks software will be used to deploy commodity Linux clusters at BIRN labs. The SDSC Storage Resource Broker software will play a key role in establishing and maintaining brain data collections. And to disseminate SDSC's networking expertise, staff experts will be visiting each BIRN site to help get the project off to a running start. The infrastructure proposed for BIRN was guided by Mark Ellisman, NPACI Neuroscience Thrust Leader and an outstanding team of collaborators. We believe BIRN will be a very exciting project. Supercomputing: What will SDSC's specific roles be in the new project and if you would, please discuss how BIRN may benefit from the TeraGrid in the future. BERMAN: SDSC staff are heavily involved in all aspects of BIRN. The BIRN Network Operations Center will be deployed at SDSC, which has a direct connection to Abilene, and SDSC will ensure that the BIRN network remains robust. SDSC experts will help each BIRN site deploy the essential network and cluster hardware configuration and provide ongoing training. SDSC researchers have developed key components of the software including NPACI Rocks cluster configuration software (developed in partnership with UC Berkeley), the Storage Resource Broker, visualization and data integration software, and other components. With respect to TeraGrid, BIRN has been identified as one of the national-scale initiatives that will be able to take advantage of the data and compute capability that the TeraGrid will make available. In addition to managing large-scale data sets, BIRN researchers may require the multi-teraflops compute power of the TeraGrid to perform the most complex comparisons and analyses on the highest-resolution brain imaging data sets. Supercomputing: On a more personal level, how big an impact do you see BIRN having on research into such diseases as depression, alzheimers, MS etc.? BERMAN: This is best answered by Mark Ellisman, a neuroscientist and BIRN's lead. It is clear however that the deployment of a high-end and resource-rich computation and data management infrastructure is fundamental for analyzing a brain data collection of the size and biodiversity critical to the discovery of new neurological results. Oftentimes, critical phenomena are only apparent with comprehensive data collections and/or at very large problem sizes. BIRN will provide a way to assemble and analyze such collections with a focus on important neurological disorders. Such discoveries are critical to advances in health and medicine over the next decade. ---------- Supercomputing Online wishes to thank Fran Berman for her time and insights. ----------
Supercomputing: Please tell us about the Biomedical Informatics Research Network (BIRN) project and how it ties in with NPACI's neuroscience & DICE thrust areas. BERMAN: The BIRN activity represents an important direction for NPACI and in particular the Neuroscience, Data-Intensive Computing Environments, and Interaction Environments thrust areas. One of the major challenges in BIRN research is data integration. Brain data sets from different labs, collected with different equipment, require incredibly sophisticated software and algorithms for permitting researchers to make comparisons among several brains. The techniques for doing so were first prototyped as part of a Neuroscience-DICE project on federating brain data. In addition, the size of the data sets presents a visualization challenge, and the scalable visualization toolkits being developed by the Interaction Environments thrust will form common infrastructure for the BIRN participants. BIRN also represents a practical application of NPACI's infrastructure building efforts. NPACI Rocks software will be used to deploy commodity Linux clusters at BIRN labs. The SDSC Storage Resource Broker software will play a key role in establishing and maintaining brain data collections. And to disseminate SDSC's networking expertise, staff experts will be visiting each BIRN site to help get the project off to a running start. The infrastructure proposed for BIRN was guided by Mark Ellisman, NPACI Neuroscience Thrust Leader and an outstanding team of collaborators. We believe BIRN will be a very exciting project. Supercomputing: What will SDSC's specific roles be in the new project and if you would, please discuss how BIRN may benefit from the TeraGrid in the future. BERMAN: SDSC staff are heavily involved in all aspects of BIRN. The BIRN Network Operations Center will be deployed at SDSC, which has a direct connection to Abilene, and SDSC will ensure that the BIRN network remains robust. SDSC experts will help each BIRN site deploy the essential network and cluster hardware configuration and provide ongoing training. SDSC researchers have developed key components of the software including NPACI Rocks cluster configuration software (developed in partnership with UC Berkeley), the Storage Resource Broker, visualization and data integration software, and other components. With respect to TeraGrid, BIRN has been identified as one of the national-scale initiatives that will be able to take advantage of the data and compute capability that the TeraGrid will make available. In addition to managing large-scale data sets, BIRN researchers may require the multi-teraflops compute power of the TeraGrid to perform the most complex comparisons and analyses on the highest-resolution brain imaging data sets. Supercomputing: On a more personal level, how big an impact do you see BIRN having on research into such diseases as depression, alzheimers, MS etc.? BERMAN: This is best answered by Mark Ellisman, a neuroscientist and BIRN's lead. It is clear however that the deployment of a high-end and resource-rich computation and data management infrastructure is fundamental for analyzing a brain data collection of the size and biodiversity critical to the discovery of new neurological results. Oftentimes, critical phenomena are only apparent with comprehensive data collections and/or at very large problem sizes. BIRN will provide a way to assemble and analyze such collections with a focus on important neurological disorders. Such discoveries are critical to advances in health and medicine over the next decade. ---------- Supercomputing Online wishes to thank Fran Berman for her time and insights. ----------