Computational infrastructure for biomedically-driven simulation and modeling Summary Computation and modeling have become integral to many biochemical, biophysical, structural, and synthetic studies. Simulations direct and inform our understanding of biomedically important processes, including RNA dynamics and catalysis, transcriptional processing, the structure and dynamics of biomolecular machines such as the ribosome and other key machinery of cellular life processes. Similarly, computations using quantum chemistry and machine learning have emerged as strong predictors of synthetic routes, reaction conditions, and catalysts. These facets of computation—understanding biochemical processes and creating pharmaceuticals which interact with them—come together within the community of molecular biophysics and chemists affiliated with this request for computational resources. Driving these developments are critical biomedical and biological questions and systems under investigation in the laboratories with whom the PI and co-Is collaborate. This cohort has made breakthroughs in many of the problems and areas noted above. Key to this success has been the close working relationship between the modeling and experimental efforts, with each feeding into one another. For more than a decade, our efforts have been supported by a computational facility as part of the “Lighthouse” computer center at the University of Michigan. To continue and grow our fruitful and productive collaborations, we propose to revolutionize the computational facility with next- generation equipment, based on computer architectures and software algorithms that were not available until recently. Today’s software, including that being developed and used by the researchers represented here, has embraced advances in GPU hardware to enable new kinds of computational studies. To make the best use of these powerful research tools, we are requesting an integrated compute server including 360 Intel Haswell CPU cores, 640 AMD 7543 cores 90 NVIDIA RTX3080Ti GPUs and 10 NVIDIA A100 GPUs as well as an associated 640 terabytes of storage and Infiniband network interconnect. This facility will be supported through the “Lighthouse” center in the University of Michigan’s research computing facility and maintained by a professional Linux cluster manager. This facility will serve a diverse core of research groups, all with established collaborative efforts and most with NIH-funded projects. Altogether, the synergistic research directions enabled by GPU computations will enable breakthroughs in biomedical sciences by the University of Michigan’s Chemistry and Biophysics research teams.

Project Narrative Recent changes in computer architectures have enabled powerful new forms of biomedically- driven simulation and modeling. Combining hardware developments—especially GPU acceleration—with software algorithmic enhancements, the detail and accuracy of computer modeling studies of biophysical, biochemical and synthetic processes have increased enormously and led to breakthroughs in protein folding, biological and abiological catalysis, transcriptional regulation, drug discovery, chemical reactivity, and structure determination. The current proposal is directed to the acquisition of a powerful 1,000 core CPU / 1,000,000 core GPU computational cluster that will enable new directions in modeling and simulation at the University of Michigan’s chemistry/biophysics research cluster.