A.5. Specific Aim 3: Multiscale Data/Output Organization and Visualization

A large part of the Center’s activities will be dedicated to developing methodology and tools for the biomedical community. However an important goal for dissemination and utilization will be to educate (see § B.3) current and future users of the methods and tools developed in the Center. To this aim, it is essential to organize and visualize the data and output from our models and methods. Three participants of the pre-NPEBC have already specialized in the development of web-accessible tools for modeling and simulating complex processes: J. Madura at the simulation, visualization and manipulation of large macromolecular systems; J.R. Stiles at the space-dependent stochastic simulations of microphysiological processes, and B. Ermentrout, at the mathematical modeling and simulation of the nonlinear dynamics of complex networks.

MCell simulations (§A.3.2) require specialized methods for model design and analysis of results, much of which is critically dependent on interactive visualization. Because the models can be very large (easily millions of molecule locations and polygons for surface meshes), MCell development has been absolutely dependent on concurrent development and optimization of underlying data structure design and visualization tools. In addition, it has also been necessary to develop a variety of interconversion tools for files generated or read by different solid modeling, mesh editing, or 3-D rendering software. Our microphysiological data structures, simulations, and visualizations now take full advantage of a powerful multidimensional data storage, analysis, programming, and rendering environment designed for and incorporated into IBM DataExplorer (now OpenDX or DX, www.opendx.org, open source). In doing so, we leverage ~1 million lines of source code that underlie OpenDX per se. Custom data manipulation and rendering programs can be created within the OpenDX environment, and a recent culmination for MCell users is DReAMM (Design, Render, and Animate MCell Models) (www.mcell.psc.edu/DReAMM).

In essence, this modeling data organization is an open-ended multidimensional spreadsheet format, with user-definable dependencies between different components, and rapid large-scale data manipulations enabled through a visual programming interface. As such, it is immediately amenable to incorporation of molecular and cellular model data, and provides a common data storage, manipulation, and visualization environment. It can also be operated with a local client that accesses data and a computational server remotely, but can still take advantage of local video hardware for high-speed rendering.

Based on this underlying design, a cohesive data organization and toolset, accessible on the Internet, will be created for building, visualizing, interchanging, modifying, and simulating models, and the computational algorithms will be practical for multiscale. An automated software download and user database facility has been created for MCell and DReAMM users (www.mcell.psc.edu/download), and a similar site will be created for the Center to disseminate datasets, computational tools, and visualization tools for users in the local, national, and international scientific communities.