Dr Kisaru Liyanage1, Dr Matthew Downton1
1National Computational Infrastructure, Canberra, Australia
Biography:
Dr Kisaru Liyanage is a High Performance Computing Specialist in Life and Health Sciences at the National Computational Infrastructure (NCI), Australia. He earned his PhD in Computer Science and Engineering from UNSW Sydney, where he focused on accelerating genomics software through hardware–software co-design. His current research interests include accelerated and high-performance computing, bioinformatics, and systems-on-chip.
Abstract:
Nextflow workflows are widely used in bioinformatics, but deploying them on shared HPC clusters can be complex. Cluster environments often require fine-tuned configurations, and systems with restricted internet access necessitate time-consuming offline preparation of workflow dependencies, such as container images and offline-ready configuration files. These challenges impede workflow adoption on HPC clusters.
We present workflow-installer, a lightweight framework that can be used to centrally deploy versioned Nextflow pipelines as environment modules with configurations optimised for specific HPC clusters. An installer script created for each workflow automates tasks such as pulling workflow source code, applying patches for offline execution, retrieving Singularity container images, downloading test datasets, configuring a script for registering workflow on Seqera platform, copying cluster-specific configurations and documentation, and generating modulefiles. workflow-installer includes several helper functions for creating reproducible, robust installation scripts (e.g., caching sources, transforming container URIs).
After deployment, users can load a workflow with “module load <workflow>/<version>” and execute it completely offline with cluster-specific optimisations. Using the pre-downloaded test datasets and example commands in the documentation, users can validate installations with minimal effort. The workflow can also be registered on the Seqera platform using the preconfigured script. Users experience virtually zero setup time and benefit from improved performance due to tailored configurations.
workflow-installer lowers the barrier to using Nextflow workflows on HPC, enabling more reproducible and efficient bioinformatics analyses. By shifting workflow setup to a single module load, researchers focus on science rather than infrastructure, making HPC resources more accessible to the life sciences community.