Mapping Fitness and Free Energy Landscapes of Proteins

点击翻译按钮获取中文摘要

Abstract, Funds are requested to purchase computer equipment to support the research projects described in the parent grant NIH 1R35GM132090 “Mapping Fitness and Free Energy Landscapes of Proteins” (R.M. Levy, PI). The PI is a computational scientist with forty years of experience in computational structural biology and biophysics. The research described in the parent grant is focused on the development and application of computational methods for studying the structure, function, and dynamics of proteins. All of the projects supported by NIH 1R35GM132090 rely on access to state-of-the-art computational resources. Since 2015, advances in software and algorithms have led to a significantly greater focus on GPU implementations for machine learning and for solving problems in computational chemistry, biophysics, and structural biology. Access to the latest GPU hardware is essential for the successful completion of the research described in the parent grant. The PI requests a supplement to purchase 8 NVIDIA A100 40GB PCIe GPUs, with double the number of GPU cores and triple the bandwidth relative to the current consumer-level GPUs the PI currently has access to, and they are much better optimized for high-performance computing. The new GPUs will be connected to an existing shared computer cluster, the Center for Biophysics and Computational Biology Research Resource (CB2RR) cluster, which supports NIH funded researchers. The GPUs will be connected to the cluster using high- throughput, low overhead InfiniBand HDR 200 Gbps connections, which will be purchased with the GPUs. In addition, funds are requested to upgrade the storage system of the computer cluster with a 500TB RAID-6, as the current hard disks are now out of warranty and at the end of their life cycle. This system will excel at accelerating highly parallelized GPU-only and CPU-GPU hybrid molecular dynamics simulations and machine learning applications as described in NIH 1R35GM132090.