DBI 9870821; Gabriel, Michael; University of Illinois, Urbana Champaign
Much current knowledge about brain function is based on analyses of the firing patterns of neurons. This field is experiencing an explosion of data. Huge data sets are being amassed with new computer-based data acquisition systems and techniques for recording simultaneously from multiple neurons and brain sites. Neural modeling generates large simulated data sets that must be analyzed and compared smith experimental data. The goal of this project is to develop an information system for all aspects of neuronal data management, including storage, retrieval, analysis, visualization and sharing of experimental and simulated data. The ultimate aim is to develop a set of tools, techniques and standards that can be disseminated to help meet the needs of a large community of neuroscientists who work with neuronal data.
Neuronal recordings can be intra-, or extracellular recordings of single spikes, ensembles of neurons and field potentials. Researchers correlate neuronal activity with other time- varying signals such as stimuli and ongoing behavior. All of these data are instances of general data type, namely time-series data. The proposed information svstem will be specialized for neuronal time-series data. This focus is complementary to efforts of other groups to develop neuroscience information systems in domains such as molecular neurobiology and functional brain imaging, which primarily deal with sequence and image data.
The overarching goal of the project is to provide a system of general utility to neuroscientists. In its current scope, the project involves the laboratories of ten facultv of the Universitv of Illinois/Beckman Institute Neuronal Pattern Analysis (NPA) Group, Which can be viewed as a prototype of the larger community of neuroscientists that will benefit from this effort. In addition the project will continue to be guided by resident colleagues of the National Center for Supercomputing Applications (NCSA), and working associations have been formed with related projects and individual researchers to incorporate a broad set of data handling conventions and requirements. The basic components of the proposed svstem have either reached or are well on the way toward completion. These include: a) a database component for organized storage and efficient search and retrieval of neuronal data meeting specified criteria of interest; b) digital brain atlases, to provide high-resolution neuroanatomical images for use in entering, retrieving and analyzing time series data; c) the time series data protocol (TSDP), a standard for representation, transfer and analysis of time series data in a platform-independent manner. d) a suite of flexible and intuitive tools for data analysis and visualization, as vvrell as a programming library for development of TSDP-compliant tools which will be interoperable within the research community.
The database component has been implemented using a generalized database table schema, allowing multiple laboratories to develop custom interfaces that work with a shared database system. Digital fish and rabbit brain atlases have been assembled and an interface is being prepared to integrate the atlases with the database and analysis functions. The atlases will be supplied to several researchers, and exportation of TSDP to other projects is planned to promote full interoperability of multiple neuronal data sets and powerful time-series analysis and visualization tools. Development of TSDP is near completion and will be field tested by on-site and external researchers. Powerful commercial, public domain and custom tools have been harnessed for data analysis and visualization, and a modular programming library is being developed to render near and existing tools TSDP-compliant and therefore widely interoperable in the research community.
The system will meet the needs of neuroscientists at several levels. It can serve as a personal database management system for individual laboratories, a "workgroup" tool for exchange of data within communities and as a means to implement an Internet-accessible ÊpublicË database. Steps remaining for full operational status are: a) completion and implementation of TSDP; b) integration of digital brain atlases erith data-entry, query- development and with analysis and visualization tools; c) activation to full on-line status of NPA pilot laboratories; d) completion of the programming library for developing TSDP-compatible analysis and visualization tools. These steps will be carried out within an ongoing cycle of performance testing, evaluation and improvement.
