Pipelines

nf-core pipelines are end-to-end analysis workflows that combine multiple tools and processes to analyze biological data. These specifications define the standards for developing high-quality, reproducible pipelines that can be shared and maintained by the nf-core community.

The specifications are organized into requirements (mandatory standards all nf-core pipelines MUST follow) and recommendations (best practices pipelines SHOULD adopt where applicable). While these specifications are mandatory for pipelines contributed to the nf-core repository, they represent proven best practices for developing professional-grade Nextflow pipelines that can benefit any workflow development project.

Requirements

The following specifications define mandatory standards that all nf-core pipelines MUST follow:

• Nextflow: Pipelines MUST be built using Nextflow and adhere to DSL2 syntax standards.
• Community owned: Pipelines are owned collectively by the community with contributions welcomed from all members.
• Identity and branding: Primary development MUST occur within the nf-core GitHub organisation with appropriate branding.
• Workflow specificity: Each data or analysis type MUST have only one canonical nf-core pipeline to avoid fragmentation.
• Workflow size: Pipelines MUST have appropriate scope—comprehensive enough to be useful but focused enough to maintain.
• Workflow name: Pipeline names MUST be lowercase, use hyphens for separation, and avoid punctuation.
• Use the template: All pipelines MUST be built using the nf-core template and maintain synchronization through automated processes.
• Software license: Pipelines MUST be open source and released under the MIT license.
• Bundled documentation: Comprehensive documentation MUST be bundled with the pipeline and hosted on the nf-core website.
• Docker support: Software dependencies MUST be containerized using Docker with specific version tags.
• Continuous integration testing: Pipelines MUST implement automated CI tests to verify functionality.
• Semantic versioning: Pipelines MUST follow semantic versioning with stable release tags.
• Standardised parameters: Pipelines MUST use standardized parameter names and conventions for consistency across nf-core.
• Single command: Pipelines MUST be executable with a single command without requiring manual intervention.
• Keywords: Pipelines MUST have comprehensive documentation and appropriate GitHub repository keywords for discoverability.
• Pass lint tests: Pipelines MUST pass all nf-core pipelines lint tests without failures.
• Credits and Acknowledgements: Pipelines MUST properly acknowledge and credit prior work, tools, and contributors.
• Minimum inputs: Pipelines MUST be designed to run with minimal required input, using sensible defaults.
• Use nf-core git branches: Pipelines MUST use standardized branch structure including master, dev, and TEMPLATE.

Recommendations

The following specifications define best practices that all nf-core pipelines SHOULD follow where possible or appropriate:

• Use Bioconda: Software dependencies SHOULD be packaged using Bioconda and distributed through BioContainers for reproducibility.
• File formats: Pipelines SHOULD use community-accepted modern file formats such as CRAM for better compression and compatibility.
• Testing: Pipelines SHOULD use nf-test to verify successful completion with valid outputs using minimal test datasets.
• DOIs: Pipeline releases SHOULD be assigned digital object identifiers (DOIs) through Zenodo for citability.
• Cloud compatible: Pipelines SHOULD be tested on cloud computing environments to ensure portability.
• Publication credit: Pipeline publications SHOULD acknowledge the nf-core community and cite the nf-core framework paper.
• Build with the community: Pipeline development SHOULD occur collaboratively with the community from the beginning rather than proposing complete pipelines.
• Custom Docker images: Custom containers not available through BioContainers SHOULD be mirrored on quay.io for long-term reproducibility.