Call genetic variants with bwa and bcftools on Deadline Cloud

The variant_calling_bwa job bundle on the GitHub website finds the genetic differences between an individual and a reference genome. It aligns sequencing reads with bwa, calls variants with bcftools, and merges the results into one VCF file. The pipeline reimplements the AWS HealthOmics WDL variant-calling tutorial pipeline in Open Job Description.

The bundle demonstrates patterns that carry over to other scientific workloads:

• Software delivered as conda packages from the bioconda channel rather than built into container images.

• Fan-out over two different task parameters: alignment scatters over samples, and variant calling scatters over genome regions, with a merge step as the gather.

• A job environment that verifies every required tool is on PATH once per session, and reports the exact packages and channels to configure if any is missing.

The bundle requires a farm with a Linux fleet, a conda queue environment with channels set to conda-forge bioconda in that order, and the Deadline Cloud CLI. Download the sample data (about 950 KB), then submit:

python sample_inputs/fetch_test_data.py
deadline bundle submit .

The default parameters (2 samples, 4 regions) produce 10 tasks. The README covers bringing your own cohort, scaling the region list, and the steps a production analysis would add. For other bioscience examples, see Predict protein structures with ESMFold on Deadline Cloud, Run GROMACS molecular dynamics simulations on Deadline Cloud, and Run virtual screening with AutoDock Vina on Deadline Cloud.