ComPRePS 2.0: Enabling Massive-Scale Distributed Computing on High-Performance Computing Cluster for...

System automation and artificial intelligence (AI) have significantly impacted the healthcare sector, leading to the rapid digitization of histological data into gigapixel resolution whole-slide images (WSI). This advancement has fueled the emergence of computational pathology, necessitating high-resolution image processing on a massive scale. To meet this need, a distributed computing environment with extreme scalability and secure architecture is essential for biomedical workloads involving large amounts of private data.. At Computational Microscopic Imaging (CMI) Laboratory, we developed and open-sourced the Computational Renal Pathology Suite (ComPRePS). This image analysis tool integrates histopathological slide visualization, quality evaluation, and AI-driven data analysis for clinical tasks. ComPRePS has already been utilized by the Kidney Precision Medicine Project (KPMP) to process tissue slides and generate data crucial for understanding kidney disease progression. The initial implementation, ComPRePS 1.0, offered a user-friendly interface and AI-driven analysis but faced limitations due to its monolithic architecture, inadequate scalability, and security concerns. These challenges prompted the development of ComPRePS 2.0, which integrates unparalleled computing power from high-performance computing clusters (HPCCs) using University of Florida's (UF) HiperGator-one of the world's largest supercomputers. By leveraging on-demand CPU, GPU and MEM resources, Apptainer-based containerization, and parallel file system access, ComPRePS 2 .0 a chieves unprecedented scalability and improved security. Benchmarking on 920 large-dimensional WSIs demonstrated a 15× performance improvement over its predecessor. ComPRePS 2.0 aims to facilitate truly large-scale automated histopathological slide analysis, benefiting researchers, pathologists, and students worldwide.