Galaxy is an open-source, web-based framework, primarily for data-intensive biomedical research, that is used by more than 20,000 researchers worldwide in many areas. Current Galaxy implementation does not support running applications on GPUs and accelerators like FPGAs. There is also no support for dynamic resource scheduling and management in Galaxy. Our objectives include re-engineering the Galaxy framework to enable GPUs and accelerators like FPGAs as "first-class" compute engines, enlarging the Galaxy community by bringing GPU and FPGA-supported tools, enabling Galaxy tools to take better advantage of emerging cluster scheduling capabilities as well as innovations on storage system management and achieving significant improvements in performance scalability and energy efficiency. We currently integrated GPU-Aware Computation Mapping for seamless execution of GPU-supported tools. We also created a containerized version of the Racon tool. Next, we created a GPU-supported version of the methylation calling tool and tested it using Galaxy. We implemented Kubernetes GPU scheduling support to enable accelerated job scheduling and execution in clusters.

NSF Award Number: OAC-1931531