BioBrowser
Prof. Dr. D. W. Fellner, PD Dr. D. W. Heinz
Computer Graphics and Knowledge Visualization,
Graz University of Technology, Austria
Ges. f. Biotechnologische Forschung Braunschweig, Germany
Motivation
Protein visualization at interactive rates has become more and more important, as the number of proteins in the RCBS Protein Data bank is increasing very fast. The ability to visualize the 3D structure of these proteins is critical in various areas such as drug design or protein modeling, because the function, which means the possible interactions with other molecules, of a protein is closely connected to its 3D structure. Therefore fast visualization tools are required, which can handle proteins with a huge number of atoms. Of course, there are visualization tools like Chimera, Cn3D, Rasmol, or FPV but they either exhibit a significant drop in rendering performance when handling very large proteins or have limited visualization styles for, e.g., ribbon structures or space fill.
Ball and Stick |
Spacefill |
Linedrawing |
|---|---|---|
| 
| 
|
Ribbon-Structure |
Solvent excluded Surface |
SES, 15x enlarged |
| 
| 
|
BioBrowser
Topology |
Surface |
Atoms |
|---|---|---|
| 
| 
|
This gap should be closed by the BioBrowser currently developed in our group. The BioBrowser is a modular framework which uses render plug-ins for the different visualization styles. Therefor it provides an easy to handle interface which makes the molecular data and any selections available to the plug-ins. This data then can be used to create a visualization of the protein. Currently the most common styles are realized as plug-ins, which are the ball-and-stick, the spacefill, the stick, the ribbon structure, and the solvent excluded surface style. The results of these renderers are shown in the images above. But through the modular structure of the framework this list can easily be extended by new visualization methods, i.e. a textual based one for the primary structure of the protein. The current implementations of the renderer made intensive use of subdivision surfaces, level-of-detail meshes, billboards and the possibilities of modern graphics cards, like fragment programs. All this is guided by the paradigm to reduce geometry complexity whenever possible and create the details on-the-fly only when really needed. This ensures high framerates when navigating through the protein, as well as high quality images, when the camera position is fixed.
Future Work
The next steps will develop new visualization styles in a coorperation with some structural biologists from the GBF, which will be implemented then. Another useful feature will be a semantic lens which will help the user to switch between the different visualization styles and to combine two or more styles in one molecule, e.g., combining a halftransparent surface with the ball-and-stick style. With all the necessary visualization features available at interactive rates, even for extremely complex proteins, we will be able to 'upgrade' the bare visualization tool into a tool which allows a researcher to embed and to retrieve additional bio-information, typically textual, as well as hyperlinks directly into the 3D molecular structures -- similar to what current web browsers offer in the textual domain. Speaking of web browsers: of course, the BioBrowser will be made available as a browser plug-in for visualizing molecules over the internet.
References
- Andreas Halm, Lars Offen, Dieter Fellner: Visualzation of Complex Molecular Ribbon Structures at Interactive Rates, Proceedings of Eighth International Converence on Information Visualisation IV04
- Andreas Halm, Lars Offen, Dieter Fellner: BioBrowser: A Framework for Fast Protein Visualization, to appear at Eurographics / IEEE VGTC Symposium on Visualization 05
This project has been funded by the German Research Foundation (DFG) (Fe431/4-3, Fe431/6-1)
