|
|
The Washington University (WU) NPACI Program is focused on applying advanced technology towards some of the difficult tasks associated with brain mapping. In particular, it is assisting other NPACI partners in constructing federated brain map databases and software tools for exploring these maps. The NPACI (National Partnership for Advanced Computational Infrastructure) is funded by the National Science Foundation (NSF) and includes approximately 46 partner sites around the world engaged in expanding the frontiers of science through the application and hardening of the most advanced computational and communication technologies. San Diego Supercomputer (SDSC), NPACI's leading-edge site, hosts the most powerful resources in the partnership, with a teraflops supercomputer and a data archive (HPSS) which will be capable of housing approximately a petabyte of data.
Washington University (WU) brings to the NPACI partnership its unique combination of world-class researchers in both Neuroscience and advanced networking. Our long-term goal is to establish a federated brain database in year 2002. The infrastructure necessary will include broadband, wide area connectivity, a universally available data storage system, acceptable use policies, specialized client and server software and brain image data sets from multiple institutions. This is a collaborative effort among the Neuroscience (NS) thrust partners.
The WU NPACI effort includes the following major activities, which are described more fully here:
The NPACI program at Washington University continues to support the Neuroscience infrastructure for the following laboratories in terms of supplying high-speed network connectivity and data cache accessibility: Mike Miller (John Hopkins University, Electrical Engineering), Marc Raichle (WU, Radiology) and David Van Essen (WU, Anatomy and Neurobiology). We plan to increase the size of the WU data cache to 1.2 TB in 1999 and to 2 TB by the end of FY 2000 to accommodate the 20 GB/week brain image data acquisition rate presently being experienced. Data acquisition activities are expected to continue to rise as the new 3 Tesla (head only) MRI scanner comes on line in the Department of Radiology in support of neuroscience activities later this year.
We plan to establish routine procedures for archiving (via the vBNS) to the HPSS located at SDSC to reduce the impact of less active data sets on our data cache. Also our data cache will be routinely backed up on the HPSS. The archiving system will require a well tested and hardened user interface which will be developed in the summer of 1999 in cooperation with the SRB (Storage Request Broker) group at SDSC. Some of this archiving experience with the SRB will be useful in making data set transfers between WU and remote NPACI sites such as JHU, UCSD, UCLA as easy and routine as local transfers.
The brain surface database and file store called SuMS (Surface Management System) developed at Washington University will be hardened and made available to other investigators by publication on the web, on the cache at Washington University and on the HPSS at SDSC. Selected brain image data sets will become available in SuMS for the use of other neuroscientists from laboratories connected to either vBNS or Abilene. Procedures for entering brain image data sets into SuMS from other institutions will be developed. A set of acceptable use policies for the SuMS and associated brain images will also be published. A generic surface visualization tool capable of display of SuMS images will be made available to the public for execution on client platforms (SGI and Sun).