The purpose of this Core C is to provide the computational expertise and resources needed for the research in this program project grant. In addition to this supporting function, the Core will perform research towards designing enhanced data acquisition and processing tools. During the current funding period we have made great progress towards optimized sampling schedules, such as the Poisson-Gap sampling, which has shown superior signal to noise and sensitivity. The FM-reconstruction together with a distill procedure has been developed and ported to computing environment using Nvidia GPU cards and is now very fast. We propose to further optimize sampling and processing methods. We realized that the distill procedure we developed is closely related to the IST (iterative soft threshold) method that has recently been reported. We have developed a new implementation of the IST approach, termed istHMS. This is very fast and allows now acquisition of 3D and 4D spectra with resolution in the indirect dimensions approaching those achievable in the direct dimensions. We have already shown that 4D methyl-methyl TROSY-NOESY experiments can be recorded with as little as 0.3 % sparsity, sampling to 118 ms in the indirect dimensions, and can be reconstructed with istHMS in less than a day. The requested upgrade of our computing cluster and code improvement will shorten reconstruction time by at least an order of magnitude. The computational innovations will have high impact on NMR spectroscopy of large proteins. We will also engage in developing new tools for automated assignments that utilize new experiments being developed here or in the NMR community elsewhere. We will pursue the following specific aims: 1. Design Optimal data acquisition strategies and educate scientists in their use 2. Develop and apply optimized and new processing methods for NUS data 3. Provide the environment and expertise with modeling, structure calculations and docking 4. Maintain the existing computing hardware and install new computers 5. Training and dissemination
Computation is key to efficient use of modern NMR spectroscopy. Optimally designed sampling and processing methods can dramatically enhance spectrometer performance by extending spectrometer resolution by more than an order of magnitude. Only with advanced acquisition and processing methods can the capabilities of modern high field instruments fully utilized. However, average users have to be trained to use these tools, and the computational resources have to be kept state of the art.
Showing the most recent 10 out of 245 publications
