"Here you can see the enzyme Trypsin. It is found in the digestive system of any vertebrate, even in your belly. The funny colors on its surface depict electrostatic potentials which are important for reactions with other substances."
The biochemist sits in front of her computer, pointing to the screen. "Now watch the small Arginin molecule moving towards the Trypsin. Do you see how it perfectly fits into the pocket on the surface? Otherwise it would not be able to operate as an inhibitor in this case. - And now, let's investigate the digestive enzyme in more detail." On the screen, you see the large molecule racing towards you. Suddenly you are inside, surrounded by huge spheres and thin cylinders. "The 'balls' you can see now", she explains, "are atoms, and the 'sticks' are bondings between them. Here you can see the atoms forming the Ca 2+ receptor..."
This short example gives us an idea of the benefit of interactive computer graphics for 'molecular modelling', e.g., design, investigation, and presentation of molecules in application domains like biology, pharmaceutics, physical chemistry, etc.
Biochemical data often are associated with a complex 3D structure. Examples are the spatial arrangement of atoms within a molecule or the electrostatic potentials on its surface. Furthermore, interaction between molecules (docking behaviour) strongly depends on these spatial relationships.
In general, data must be interpreted in order to become useful information. Most biochemical data are best interpreted by the use of 3D representations. In the past, wooden or plastic models were used in order to investigate the three-dimensional structure of molecules. Today, computer graphics allows to view molecules in 3D in a much more flexible way without the tedious process of assembling small physical parts by hand.
In biochemical applications of computer graphics, two opposing demands are:
High quality rendering has been done for many years using raytracers. This has lead to impressive animations which are being used in education or to communicate research results to a wider audience. Sometimes researches report that a high quality 3D image of their data has lead to new insights and to a better understanding of their own work.
Interactive handling of digital molecule models can further improve understanding dur to the complex 3D structure of molecular data. It has become feasible due to the tremendous advances in hard- and software. Until recently, only very schematic representations of molecules (e.g. line models or dot clouds) have been possible in real time and input devices like keyborad or dials were not very intuitive to use.
The system `VRMol` of Fraunhofer-IGD aims at the highests possible degree of rendering quality while at the same time utilizing Virtual Reality techniques in order to allow interactive investigation of individual molecules as well as of docking beaviour between different molecules, e.g., between enzyme and substratum.
Several Virtual Reality and Scientific Visualization techniques have been implemented within this system. Special emphasis was given to fast rendering of large molecules in high quality (level of detail, fast rendering of spheres, sorted polygons for transparent surfaces), on good interaction (simultaneous operation of two input devices with at least 6 degrees of freedom each, virtual menus), and on judgement of position and size of molecules (fast shadows).
Several levels of detail are used to represent atoms, bondings, and surfaces. Depending on the distance between viewer and object, a geometric representation is chosen for the object which posesses as few polygons as possible while still looking realistically. These geometric models were automatically generated with an inhouse reduction algorithm.
Ball-and-stick representations of molecules contain a large number of spheres depicting atoms. In VRMol, these spheres are not rendered as ordinary polygonal objects. Instead, a fast, specialized algorithm for rendering spheres is being utilized for higher performance.
The 'surface' of a molecule can be displayed transparently with surface colors depicting electrostatic potential, hydrophobic potential, etc. Rendering of transparent surfaces is supported by graphics hardware, but the transparent polygons need to be sorted in viewing direction in order to avoid artefacts. Thus, for each surface we precompute a number of objects sorted according to different viewing directions. During real time operation the system selects the object which gives best results according to the current viewing angle.
In order to allow simultaneous movement of viewer and a molecule, two different multidimensional input devices are supported: a data glove for viewing and a space mouse for rotating and translating a molecule. The data glove allows intuitive handling by means of gestures. Thus, it is possible to move forwards or backwards just by performing different gestures with the data glove. The space mouse, on the other hand, remains on the desk all the time. It allows more exact positioning in 6 dimensions than the glove.
Virtual menus allow selection of the molecules to be displayed as well as of their parameters: ball-and-stick representation with or without simultaneous surface model, etc. A virtual menu is activated by a certain gesture of the data glove. Similar to an ordinary 2D menu it offers a number of choices of which one can be selected. Selection is performed by moving the index finger of the data glove echo to the appropriate entry. Each item of the menu is represented by a text as well as by a colored icon for easy recognition.
Finally, fast shadows have been implemented in order to give the viewer an impression of shape, size, and position of each molecule in addition to the stereo display. Shadows have been realized by computing several standard shapes and by displaying the appropriate one as a texture on the synthetic floor of the scene.
All the techniques which were realized in this system contribute in one way or another to the final goal of investigating molecules in high quality at interactive frame rates. They may also be used in other applications like assembly tasks, virtual material testing, medical training, 3D weather forecast, etc. VRMol is an example of the fruitful synergy between Virtual Reality and Scientific Visualization.
Helmut Haase
(haase@igd.fhg.de),
Johannes Strassner
(strassne@igd.fhg.de),
Fan Dai
(dai@igd.fhg.de)
