Real –Time Ambient Occlusion for Visualization of Multi-Scale Molecular Models

Author: Manuel Wahle

Primary Advisor: Stefan Birmanns, PhD (co-author)

Committee Members:

Masters thesis, The University of Texas School of Health Information Sciences at Houston.

Abstract:

Many different experimental and computational techniques contribute to the understanding of large biomolecular complexes. Cryo-electron microscopy and tomography are examples of experimental methods that reveal the structure of biomolecules on different levels of resolution. While the techniques that explore the structure of small systems yield results on an atomic level of detail, the biomolecular shapes of larger systems can typically only be imaged as volumetric maps at a much low resolution. Consequently, it is a routinely employed technique to reconstruct a high resolution representation of large systems by docking the smaller atomic models into the large volumetric maps. Through this multi-scale modeling, the data sets are integrated and the large systems can be resolved with high resolution, too. In early times, these docking procedures were carried out visually. But the need for computational methods emerged quickly to help resolving ambiguities and to provide objective measures. Semi-interactive methods of multiscale modeling employ a combination of computational and manual techniques. Human knowledge and judgment capabilities are supported by the precise results of scoring functions. For techniques that involve interventions a high quality-visualization of the atomic structures and volumetric maps is of crucial importance. We present a method that approximates ambient occlusion, a global lighting effect that greatly supports the perception of spatial expansions. Our method is implemented into Sculptor, a mature multiscale modeling and visualization software. Sculptor incorporates a scene-graph based infrastructure to display the molecular models, leveraging different rendering techniques for efficient rendering. To obtain full compatibility with these, we implemented the ambient occlusion computation as a post-processing step, exploiting the computational power and flexibility of the graphics processor.