Instant construction of atomistic models for visualization in integrative cell biology

Abstract

Computational models have advanced research of integrative cell biology in various ways. Especially in the biological mesoscale, the scale between atoms and cellular environments, computational models improve the understanding and qualitative analysis. The mesoscale is an important range, since it represents the range of scales that are not fully accessible to a single experimental technique. Complex molecular assemblies within this scale have been visualized with x-ray crystallography, though only in isolation. Mesoscale models shows how molecules are assembled into more complex subcelluar environments that orchestrate the processes of life. The skillful combination of the results of imaging and experimental techniques provides a glimpse of the processes, which are happening here. Only recently, biologists have started to unify the various sources of information. They have begun to computationally assemble and subsequently visualize complex environments, such as viruses or bacteria. Currently, we live in an opportune time for researching integrative structural biology due to several factors. First and foremost, the wealth of data, driven through sources like online databases, makes structural information about biological entities publicly available. In addition to that, the progress of parallel processors builds the foundation to instantly construct and render large mesoscale environments in atomistic detail. Finally, new scientific advances in visualization allow the efficient rendering of complex biological phenomena with millions of structural units. In this cumulative thesis, we propose several novel techniques that facilitate the instant construction of mesoscale structures. The common methodological strategy of these techniques and insight from this thesis is “compute instead of store”. This approach eliminates the storage and memory management complexity, and enables instant changes of the constructed models. Combined, our techniques are capable of instantly constructing large-scale biological environments using the basic structural building blocks of cells. These building blocks are mainly nucleic acids, lipids, and soluble proteins. For the generation of long linear polymers formed by nucleic acids, we propose a parallel construction technique that makes use of a midpoint displacement algorithm. The efficient generation of lipid membranes is realized through a texture synthesis approach that makes use of the Wang tiling concept. For the population of soluble proteins, we present a staged algorithm, whereby each stage is processed in parallel. We have integrated the instant construction approach into a visual environment in order to improve several aspects. First, it allows immediate feedback on the created structures and the results of parameter changes. Additionally, the integration of construction in visualization builds the foundation for visualization systems that strive to construct large-scale environments on-the-fly. Lastly, it advances the qualitative analysis of biological mesoscale environments, where a multitude of synthesized models is required. In order to disseminate the physiology of biological mesoscale models, we propose a novel concept that simplifies the creation of multi-scale procedural animations.