This page shows the details of why an item matched the keywords from your search.
Property | Value |
abstract
|
Project Summary/Abstract Biomedical research depends increasingly on structural and dynamical information about the networks of macromolecular assemblies that underlie all biological function. High-performance computing (HPC), modeling, and large-scale simulations are expected to increasingly play a crucial role in this process. This situation is particularly true in the case of biomolecular systems. Our understanding begins with structural information. Recent advances in cryo-electron microscopy (cryo-EM) technology have led to a ?resolution revolution? in structural biology and is now possible to reconstruct 3D molecular structures at ?-scale resolution from the analysis of 2D cryo-EM images. However, the treatment of cryo-EM data leading to the determination of molecular structure at atomic resolution is computationally intensive. Ultimately, a genuine mechanistic understanding of complex biomolecular systems will be recognized by the ability to make accurate quantitative predictions of structure, dynamics and function from computational models. In particular, the ability to create a virtual reality through classical molecular dynamics (MD) simulations has now become an integral part of the investigation process in biomedical research. Our NIH-funded research projects rely on intensive computational analysis. The acquisition of the Beagle supercomputer in 2010 (and its upgrade in 2015), which was made possible by grants from the NIH, National Center for Research Resources (NCRR), has spurred the emergence of a vibrant computational community across UChicago and the wider Chicago region. We propose to continue the Beagle project?s success by expanding into a 3rd phase that addresses the critical HPC needs of fundamental NIH-funded research in biomolecular structural and computation. Experience shows that HPC is achieved by a combination of strong CPU (Central Processing Unit) and GPU (Graphics Processing Unit). Pioneered in 2007 by NVIDIA, a GPU has a massively parallel architecture consisting of thousands of small but efficient processors designed to handle multiple tasks simultaneously. GPU-accelerated computing is now commonly used for a range of scientific computations in science and engineering;8 it achieves unprecedented performance by offloading compute-intensive portions of the application to the GPU, while the remainder of the code runs on the CPU. The need to meet our scientific objectives and the very large demand for high performance computing resources on campus to support NIH-funded research projects justifies this request for a state-of-the-art GPU cluster. The proposed Beagle-3 ? A Shared GPU Cluster for Bimolecular Science ? will create new synergies between a multidisciplinary group of users, enable quantitative assessment and validation of biomolecular models and significantly increase efficiency and productivity.
|