Advances in instrumentation have caused an exponential explosion in the amount of data generated for biomedical research, of which imaging is a prime example. A single researcher today can generate up to 15 Gb of high resolution image data in a single afternoon. In this situation, a typical lab with multiple users and studies can result in up to 5 to 20 terabytes per year. The manual processing of a study is time consuming and can require the equivalent of 2 FTE (full time equivalent) years of a scientist's time. The bottleneck for scientific progress is the laborious manual processing and analysis of the data. This proposal contends that the Pi's commercially successful data management system and innovative automated workflow technologies (created with open source web-based platform technologies) can be applied to create a robust and affordable Collaborative Analysis Management System ("CAMS"). This system will provide the ability to: 1) Easily take an existing set of labor-intensive steps and translate those steps into a defined executable script or workflow, 2) Easily enable the workflow technology to automatically execute the steps on specified data, 3) Easily send intensive processing tasks to available local or remote computing resources, 4) Easily manage the raw data, resulting processed data and history of scientific workflow in an integrated secure system accessible via the Internet, and 5) Easily allow labs to share reproducible workflows and results. This enabling technology would have the following impact on scientific practice: a) Significantly increase the progress rate of a research study by potentially eliminating FTE-years of manual effort per study, b) Increase the accuracy and repeatability of analysis and results, c) Increase the ability of researchers to leverage other's methodologies and raw data, and d) Increase the ability to verify past research study and analysis. The PI proposes to focus on imaging as a representative data type commonly found in academic and clinical environments. Phase 1 will extensively use the expertise of key researchers from the San Diego Supercomputer Center (SDSC) at DC San Diego, Caltech, UCSD, UCLA, UCSF and Lawrence Berkeley Laboratories for prototype development and testing
