<![CDATA[Efficient resource utilization is a critical challenge in High-Performance Computing (HPC) environments, particularly for long-read genome assembly workflows that require substantial computational resources. This study presents an empirical benchmarking framework to optimize resource allocation for de novo long-read genome assembly of Acacia crassicarpa. Nine experimental scenarios were evaluated by varying CPU cores (32, 48, and 64) and memory allocations (32 GB, 64 GB, and 128 GB) managed via the Slurm workload manager. Performance was assessed based on execution time, assembly continuity (N50), and biological completeness using BUSCO. The results demonstrate that CPU scalability significantly impacts performance, reducing execution time by up to 49% when scaling from 32 to 64 cores. Conversely, increasing memory allocation beyond 64 GB yielded no significant improvements in assembly quality, highlighting the risks of resource over-provisioning. Scenario 2 (64 CPU cores and 64 GB RAM) was selected as the optimal configuration because it balanced runtime, N50 continuity, memory efficiency, and BUSCO completeness, not because it produced the absolute shortest runtime. Under Scenario 2, the workflow achieved an average runtime of 59 hours 39 minutes 40 seconds, an N50 value of 7.8 Mb, and a genome completeness score of 99.8%. These findings provide practical guidance for resource planning and workload scheduling in shared HPC-based genomic workflows.]]>
Copyrights © 2026
