1. C omputer G raphics, TU Braunschweig EuroVis 20051 “ BioBrowser: A Framework for Fast Protein Visualization ” Andreas Halm, Lars Offen, Dieter Fellner (TU Braunschweig) IEEE VGTC Symposium on Visualization (2005)

2. C omputer G raphics, TU Braunschweig EuroVis 20052 Overview  Existing tools  Motivation  BioBrowser  Overall concepts  Different visualization styles  Future work  Live Demo

3. C omputer G raphics, TU Braunschweig EuroVis 20053 Existing Tools  Visualization Tools  RasMol, Chimera, PyMol, Yasara, and others  Modeling Tools  VMD (Visual Molecular Dynamics), NAMD (Scalable Molecular Dynamics), ProtoShop (Interactive Protein Manipulation), etc.  Database Tools  BLAST, FASTA (search for similar structures)

4. C omputer G raphics, TU Braunschweig EuroVis 20054 Motivation  Increasing number of structurally fully analyzed molecules  Many different sources of information, e.g.  Protein database  UniProt  GenomeNet  Provide a tool, which integrates this information into the 3D structure of the molecule … (where it really belongs)

5. C omputer G raphics, TU Braunschweig EuroVis 20055 BioBrowser - Goals  Provide an integrated access and visualization tool for existing information sources  Challenges:  Visualization (make access as smooth as possible)  Support all common visualization styles  Interactive presentation of huge datasets  Keep precomputed data to a minimum  Integration of the information  Interaction with existing tools for data collection  Integrate collected information into 3D space

6. C omputer G raphics, TU Braunschweig EuroVis 20056 Overall Concepts I  Visualization of huge data sets in real time  Just-in-Time generation of the geometry  Reduction of geometry through  Level-of-Detail  Billboarding  Subdivision-Approaches  Use of modern GPU Features  Minimizing precomputed vis-related data  Designing/Use of fast algorithms  Data (e.g. solvent surface geometry) can be made available in real time

7. C omputer G raphics, TU Braunschweig EuroVis 20057 Overall Concepts II  Ensure Extensibility  Compact Interface for plugins to integrate new modules (which also supports)  Communication between the different modules  Platform independent  No usage of platform-dependent libraries, like MFC  Use of wxWidgets  Version for: Windows, Linux, Cave, Mozilla/IE- Plugin, PDA (work in progress)

8. C omputer G raphics, TU Braunschweig EuroVis 20058 Visualization Styles

9. C omputer G raphics, TU Braunschweig EuroVis 20059 Visualization Styles

10. C omputer G raphics, TU Braunschweig EuroVis 200510 Different approaches Decreasing Geometry Decreasing GPU functions LOD-Meshes Modeled polytopes Subdivision Surfaces Adding depth Texture =>Depth sprites Fully GPU rendered Ray-shooting on the GPU [Klein&Ertl 2004] Billboards

11. C omputer G raphics, TU Braunschweig EuroVis 200511 Ball and Stick  Spheres don’t intersect and the bonds intersect them in a predictable manner => use billboards, results in four vertices per billboard  If graphics board is capable of vertex/fragment shaders use a ray-shooting shader => only one vertex per atom or bond

12. C omputer G raphics, TU Braunschweig EuroVis 200512 Spacefill  Using van der Waals radius for atoms  Different approaches (GPU capability)  LOD-Meshes  Depth sprites  Fully GPU rendered

13. C omputer G raphics, TU Braunschweig EuroVis 200513 Different approaches LOD-Meshes Modeled polytopes Depth sprites Billboards with depth texture Fully GPU rendered Ray-shooting on the GPU Decreasing Geometry Decreasing GPU functions

14. C omputer G raphics, TU Braunschweig EuroVis 200514 Ribbons  Represents the structure of the backbone  C α -atoms used for position and appearance  Three different kinds:  Helices,  -Strands, turns

15. C omputer G raphics, TU Braunschweig EuroVis 200515 Ribbons – Technique  Coarse base mesh  Build quads  Subdivision using combined BReps (polygonal faces + Catmull/Clark)  Fine tuning of subdivision level  Curvature, projected size, idle time

16. C omputer G raphics, TU Braunschweig EuroVis 200516 Surfaces  What are molecular surfaces?  Instead of using fixed resolution (like Marching Cubes), use subdivision surfaces  Leads to both:  high quality visualization  interactivity

17. C omputer G raphics, TU Braunschweig EuroVis 200517 Surfaces – Calculation  Calculation:  Molecule  Spacefill  Reduced Surfaces  Base mesh  Subdivision surfaces

18. C omputer G raphics, TU Braunschweig EuroVis 200518 Structural Lens  Combine different visualization styles using their respective technique  Improve understanding which substructures define the surface and, therefore, determine the molecule’s function

19. C omputer G raphics, TU Braunschweig EuroVis 200519 Future Work  Integrate common tools (e.g. as plug-ins)  Sequence-analysis/alignment  Blast  FASTA  Structure definition  DSSP  Embed additional bio-information, collected from internet or local databases into 3D- structure

20. C omputer G raphics, TU Braunschweig EuroVis 200520 Credit The BioBrowser results from a collaboration project (funded by the German Research Foundation, DFG) between the  Institute of Computer Graphics, TU Braunschweig and  Structural Biology, Ges. f. Biotechnologische Forschung (GBF), D. Heinz, G. Dieterich, J. Reichelt

21. C omputer G raphics, TU Braunschweig EuroVis 200521 Live Demo Demo

22. C omputer G raphics, TU Braunschweig EuroVis 200522 References  Halm A., Offen L., Fellner D.: Visualization of Complex Molecular Ribbon Structures at Interactive Rates. Proceedings of the Information Visualisation (IV’04), pp. 737-744  Klein T., Ertl T.: Illustrating Magnetic Field Lines using a Discrete Particle Model. VMV 2004

23. C omputer G raphics, TU Braunschweig EuroVis 200523 Thank you for your attention!