A high performance computing cluster to support structural biology

Project Summary/Abstract (Description) The aim of this Shared Instrumentation Grant application is to provide the growing structural biology community at UT Health San Antonio (UTHSA) with access to High-Performance Computing capabilities. The UTHSA Structural Biology Core Facility (SBC) enjoys a rich 25-year history in structural biology with a

consistent group of NIH-funded researchers using structural techniques (cryo-EM, X-ray crystallography, and NMR) as a hallmark of their research programs focused on drug discovery and basic mechanisms aiming to improve human health. While the SBC has managed to provide users with adequate computational and

storage resources to process X-ray crystallography and NMR data over the years, a severe computational bottleneck for SBC users has arisen due to: 1) the recent addition of cryo-EM to the suite of structural biology tools available at UTHSA and 2) the emergence of AI-based software used to process experimental structural

biology data and for modeling of large protein complexes. Both of these use cases require powerful hardware that is currently unavailable anywhere on campus and is cost prohibitive for most individual investigators, thereby creating a barrier to entry that limits the impact the SBC has on its users’ research programs.

These computing needs of the SBC are best served by dedicated hardware specifically tailored to meet the system requirements of workhorse software used for cryo-EM, X-ray crystallography, and NMR data processing. The proposed equipment prioritizes shared memory, high-throughput architecture, and storage

and consists of a GPU cluster of 6 nodes with 24 RTX A6000 GPUs with 48GB RAM in addition to a single node with 4 state-of-the art A100 GPUs with 80 GB RAM. Each node comes with 2x Intel Xeon scalable 16- core processors. For further parallelization and for less computation-heavy processes, a CPU cluster of 4x

Intel Xeon Scalable Gold 6330H Processor 24-Cores is included along with 1.5 petabytes of raw storage capacity. This state-of-the-art configuration will allow SBC users to tap into all of the latest software for structural biology data processing and is particularly well-suited for computationally intensive cryo-EM and AI-

based data processing that is sorely needed on our campus. The proposed High-Performance Computing Cluster will be housed in the University’s Advanced Data Center and will provide cost-effective resources for 15-20 UTHSA SBC users to simultaneously process cryo-EM, X- ray, and NMR data using current state-of-the-art software while also opening up AI-based processing capability

to an even larger cohort of investigators on campus. The cluster will provide a boost to the productivity and accessibility of our new cryo-EM laboratory at a crucial time in its development, while extending the capabilities of longstanding SBC infrastructure. From a broader perspective, the cluster will accelerate the research efforts

of NIH-funded investigators, enrich the intellectual and scientific environment at UTHSA, foster development of joint grant applications, and create new training opportunities for postdoctoral and graduate trainees.