Research Interests: Michelle Kuttel
I am looking for MSc and PhD students for interdisciplinary research projects in the fields of simulation, scientific visualization
and parallel computing, as outlined below.
Project's 1 and 2: Implementation parallel algorithms on the GPU architecture
Project 3: Developing New Molecular Visualizations
I have a long-standing interest in parallel and multi-core computing. Hybrid architectures combining multicore CPUs with GPU accelerators have recently shown great potential applications in the areas of simulation and visualization. We have a number of projects in this area. Currently, Ian Tunbridge is working on an MSc project involving protein simulations on GPU's. This project will soon complete, but there is considerable potential for extending the work to support more complex simulations. The second project would look specifically at the implementation of algorithms for handling large graphs, which have broad applicability, on the GPU architecture.
Project 4: Protein-protein interface prediction
Molecular structures essentially consist of atoms ("balls") linked by bonds. Molecular visualization aims to represent molecular structures in ways that highlight important chemical and structural features of the molecule. The right visualizations can provide amazing insights into molecular structure and function.
Since 2004, we have been working on the development of novel visualization algorithms for biological molecules. This work is now incorporated into the molecular visualization package VMD (http://www.ks.uiuc.edu/Research/vmd/) and has been published. However, much scope remains for developing, implementing and testing new visualizations.
Requirements: Very little knowledge of chemistry is required for this project: molecular structure files simply list atom types and their xyz location in space. What is required is a desire to develop new and innovative visualizations. As this project has a large graphical component, James Gain and I would both supervisors. Students would obtain skills in scientific visualization and graphics, as well as skills in programming for Open Source software and, of course, so knowledge of biological molecular structures.
Protein-protein interactions play vitally important roles in many biological processes. Proteins interact at specific sites on their surface. Locating these sites is very important for a variety of reasons, from understanding the mechanism of protein interaction, to the design of new drugs that interfere with these interactions to prevent, for example, virsuses from assembling. This project would build on an tool, Piech, developed by MSc student Marco Gallotta. Piech is a tool for characterizing protein interfaces. This project would use Piech to develop a predictive tool that would attempt to locate the site of protein-protein binding interfaces in a protein structure. As with Project 3, this project does not require a knowledge of chemitry/biochemistry - the requisite background can be obtained during the MSc.
Other projects: Software for Astronomy
We (myself and Patrick Marais) have recently set up a collaboration with the Astronomy department at UCT. There are a number of interesting MSc projects for developing software tools and novel algorithms that can be pursued in this area.
Other projects: Particle Simulations
Simulations of particles interacting under the influences of force equations have broad applications. I originally pursued particle simulations as molecular dynamics simulations of carbohydrate molecules. These studies are still underway, but collaboration with Drs Gain and Marais has broadened the scope to include particle simulations of environmental phenomena (water, hair, sand) of use to the visual effects and gaming industries. Particle simulations are typically very resource intensive and are thus ideal candidates for parallelization.
Potential students interested in these project areas are encouraged to contact me (firstname.lastname@example.org) to discuss the details.