WDL workflow for short-read variant calling using deepvariant in capture-based panel sequencing.
WDL workflow for short-read variant calling using DeepVariant. It is designed to be modular, reproducible, and optimized for use in clinical settings.
- WDL (Workflow Description Language)
- Cromwell (Workflow Execution Engine)
- Apptainer (for containerized tools)
This workflow is designed to be used on HPC cluster with apptainer images. You could install images from this repo : https://github.com/MobiDL/apptainer-recipes
- Sam/Bam/Cram file(s)
- Reference genome (e.g., GRCh38)
- Target regions (BED file for panel definition)
git clone --recursive https://github.com/MobiDL/DeepVariant.git
cd DeepVariantBefore running the workflow, adapt the backend configuration to match your HPC environment.
Common parameters to review include:
queuetmp_dirroot_dirtemporary-directoryroot
These parameters are usually cluster-specific and may need to be adjusted depending on your scheduler and storage architecture.
A minimal test dataset is available in the tests directory.
Update the paths in tests/test.json, then run:
java cromwell run alignmentSR.wdl \
-Dconfig.file=backends.conf/slurm_apptainer.conf \
-i tests/test.jsonEdit the inputs.json file to specify your input files and parameters:
{
"DeepVariant.reads": "path/to/sample1.bam"
"DeepVariant.reference": "path/to/reference.fa",
"DeepVariant.target_regions": "path/to/targets.bed"
}java cromwell run alignmentSR.wdl -Dconfig.file=backends.conf/slurm_apptainer.conf -i inputs.jsonDeepVariant/
├── backends.conf/ # Backends sub-repository
├── modules/ # Modules sub-repository
├── tests/ # tests directory containing minimal dataset
├── DeepVariant.wdl # Main workflow file
└── README.md # This file
- VCF files: Variant Calling Format file(s)
Contributions are welcome! Please open an issue or submit a pull request for any improvements or bug fixes.
For any questions or issues, please open an issue in this repository or contact us.