Executions of CWL-based workflows along the cloud continuum on serverless platforms

Computational workflows are fundamental in data-driven fields such as bioinformatics and require automation, reproducibility, and scalability. Different workflows, or tasks inside the workflow, may require different resources or data backends, benefiting from executing across the cloud continuum. The execution of these workflows is feasible through serverless platforms that support deployment and execution across the cloud continuum, such as OSCAR. However, the current platform does not yet support the orchestration, traceability, and flexible execution of computational workflows. Here we present CWL-OSCAR, a new workflow management system that orchestrates computational workflows written in Common Workflow Language on the OSCAR platform. We position CWL-OSCAR within the landscape of existing serverless workflow systems and evaluate it against the reference CWL runner on the same constrained hardware. Beyond its general applicability, we emphasise our solution’s value in resource-limited environments by demonstrating its feasibility through the execution of a bioinformatics workflow on mobile genomic surveillance that runs on constrained edge devices and scales seamlessly to cloud resources when available.