WILDS WDL Library A centralized collection of bioinformatics WDL infrastructure providing reusable, well-tested components that can be combined to create powerful computational pipelines for genomics research.

Overview

The WILDS WDL Library consolidates bioinformatics workflows into a single, well-organized repository that serves as both a collection of production-ready tools and a demonstration of WDL best practices. Rather than maintaining separate repositories for each workflow, this library promotes modularity and reusability through a two-tier architecture.

Library Architecture

The library is organized into two complementary levels:

Modules (modules/)

Tool-specific collections of reusable WDL tasks with comprehensive testing.

• Purpose: Foundational building blocks for larger workflows
• Content: Individual bioinformatics tools (STAR, BWA, GATK, etc.)
• Testing: Unit tests ensure each task functions correctly over time
• Usage: Import tasks into custom workflows or run demonstration workflows

Pipelines (pipelines/)

Complete analysis workflows combining multiple modules.

• Purpose: Functional pipelines ranging from educational examples to production-ready analyses
• Content: Multiple modules combined into analysis workflows of varying complexity
• Complexity Levels: Basic (2-3 modules), Intermediate (4-6 modules), Advanced (10+ modules)
• Testing: Integration tests verify modules work together seamlessly
• Usage: Templates for common workflows, learning examples, or production analyses

Quick Start

Running Pipelines Directly (No Clone Required)

Thanks to GitHub URL imports, you can download and run any pipeline without cloning the entire repository:

# Download a pipeline and its example inputs
# Option 1: Use curl from the command line
curl -O https://raw.githubusercontent.com/getwilds/wilds-wdl-library/main/pipelines/ww-sra-star/ww-sra-star.wdl
curl -O https://raw.githubusercontent.com/getwilds/wilds-wdl-library/main/pipelines/ww-sra-star/inputs.json
# Option 2: Download directly from GitHub by navigating to the file and clicking the download button

# Modify inputs.json as necessary for your data, then run via the command line or PROOF's point-and-click interface
sprocket run ww-sra-star.wdl inputs.json

Using the Full Repository

If you want to explore multiple components or contribute:

# Clone the repository
git clone https://github.com/getwilds/wilds-wdl-library.git
cd wilds-wdl-library

# Run a module test workflow (no inputs needed)
cd modules/ww-star
sprocket run testrun.wdl

# Run a pipeline (modify inputs.json as necessary)
cd ../../pipelines/ww-sra-star
sprocket run ww-sra-star.wdl inputs.json

Importing into Your Workflows

import "https://raw.githubusercontent.com/getwilds/wilds-wdl-library/refs/heads/main/modules/ww-sra/ww-sra.wdl" as sra_tasks
import "https://raw.githubusercontent.com/getwilds/wilds-wdl-library/refs/heads/main/modules/ww-star/ww-star.wdl" as star_tasks

workflow my_analysis {
  call sra_tasks.fastqdump { input: sra_id = "SRR12345678" }
  call star_tasks.star_align_two_pass {
    input: sample_data = { "name": "sample1", "r1": fastqdump.r1_end, "r2": fastqdump.r2_end }
  }
}

WILDS pipelines use GitHub URLs for imports, providing several advantages:

• No local cloning required: Use modules directly without downloading the repository
• Version control: Pin to specific commits or tags for reproducibility
• Easy updates: Switch between versions by changing the URL
• Modular usage: Import only the modules you need

Supported Executors

All components are tested with multiple WDL executors:

• Sprocket: Modern WDL executor with enhanced features
• Cromwell: Production-grade workflow engine
• miniWDL: Lightweight local execution

For Fred Hutch Users

Fred Hutch researchers can use PROOF to submit workflows directly to the on-premise HPC cluster. This provides a user-friendly interface for researchers unfamiliar with command-line tools while leveraging the power of the institutional computing resources.

Cromwell Configuration: PROOF users can customize workflow execution using Cromwell options. See cromwell-options.json for example configurations including call caching, output directories, and more. For detailed information, refer to the Cromwell workflow options documentation.

Platform-Specific Configurations: Some pipelines include optional platform-specific configurations (e.g., .cirro/ directories) for execution on cloud platforms like Cirro. These configurations are self-contained within each pipeline directory.

Quality Assurance

Automated Testing

• Continuous Integration: All components tested on every pull request
• Multi-Executor Validation: Ensures compatibility across different WDL engines
• Real Data Testing: Uses authentic bioinformatics datasets for validation
• Scheduled Monitoring: Weekly checks detect infrastructure changes
• HPC Monthly Test Runs: Resource-intensive modules excluded from GitHub CI are validated monthly on the Fred Hutch HPC, with results posted to a tracking issue for visibility

Standards and Best Practices

• Standardized Structure: Consistent organization across all components
• Container Management: Versioned, tested Docker images from the WILDS Docker Library
• Documentation Standards: Comprehensive README files and inline documentation
• Version Control: Semantic versioning and careful dependency management

AI Disclosure

Large language models (primarily Claude) have been used to assist with prototyping new modules and pipelines in this repository, including initial drafts of task scaffolding, command structures, and test workflows. All AI-generated code is reviewed, validated, and modified as necessary by human researchers through linting tools, multi-executor test runs (Cromwell, miniWDL, Sprocket), and peer code review. No architectural or scientific decisions are made by AI tools.

Contributing

We welcome contributions at all levels:

Adding New Modules

1. Focus on high-utility bioinformatics tools
2. Follow the standard module structure
3. Include comprehensive tests and validation
4. Provide detailed documentation

Creating Pipelines

1. Combine existing modules (prefer existing modules over new tasks)
2. Demonstrate common analysis patterns
3. Include realistic test datasets
4. Document complexity level and integration approaches

Improving Documentation

• Enhance existing README files
• Add usage examples and tutorials
• Improve inline code documentation
• Contribute to the WILDS documentation site

See our Contributing Guidelines for detailed information.

Development Roadmap

Current Focus

• Growing the library through the WILDS WDL Development Program, a collaborative initiative where Fred Hutch researchers partner with the OCDO Research Informatics team to convert their computational analyses into production-quality WDL workflows
• Expanding module and pipeline coverage based on researcher needs identified through the development program
• Enhancing cloud platform integration (e.g., Cirro) for broader execution options

Future Plans

• Continued researcher-driven development of modules and pipelines through the WILDS WDL Development Program
• Additional advanced pipelines for publication-ready analyses
• Enhanced documentation, tutorials, and community outreach

Support

• Issues and Bug Reports: GitHub Issues
• General Questions: Contact the Fred Hutch Office of the Chief Data Officer (OCDO) at wilds@fredhutch.org
• Documentation: Contributing Guidelines
• Fred Hutch Users: Scientific Computing Wiki

Citation

If you use the WILDS WDL Library in your research, please cite it. The DOI below always resolves to the latest release; see the Zenodo record for version-specific DOIs.

Firman, T., Bishop, E., et al. (2026). WILDS WDL Library [Computer software]. GitHub. https://doi.org/10.5281/zenodo.19339693

BibTeX:

@software{wilds_wdl_library,
  author    = {Firman, Taylor and Bishop, Emma and others},
  title     = {WILDS WDL Library},
  year      = {2026},
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.19339693},
  url       = {https://doi.org/10.5281/zenodo.19339693}
}

A machine-readable citation is also available in CITATION.cff, which GitHub uses to power the "Cite this repository" button in the sidebar.

Dockstore

All modules and pipelines in this library are published on Dockstore, enabling discovery, import, and citation of individual workflows. Browse the library on Dockstore to find modules and pipelines with detailed descriptions, author information, and DOIs.

Related Resources

• WILDS Docker Library: Container images used by WDL workflows
• WILDS Documentation: Comprehensive guides and best practices
• Fred Hutch SciWiki: Institutional computing resources and tutorials

License

Distributed under the MIT License. See LICENSE for details.