vargate 1.0.0

Context-aware QC scoring for variant calling (Picard + biobambam2)

Turn post-alignment QC metrics into a PASS / WARN / FAIL verdict you can act on

---

VarGate reads post-alignment QC metrics (Picard + biobambam2 markdup) for one tumor/normal pair and returns an actionable verdict (PASS / WARN / FAIL), calibrated via a YAML profile for the downstream analysis.

It was built for somatic variant calling and defaults to that workflow, but YAML profiles allow thresholds and rules to be customized for other workflows. It processes one patient at a time and fits into Snakemake loops over wildcards.

VarGate ingests Picard CollectWgsMetrics, CollectAlignmentSummaryMetrics, CollectInsertSizeMetrics, and optionally CollectGcBiasMetrics, plus biobambam2 bammarkduplicates2, and writes a self-contained HTML report and a TSV summary.

Custom profiles can be added under src/vargate/profiles/.

Installation

pip install vargate

Or from source:

git clone https://github.com/pieetie/vargate
cd vargate
pip install -e ".[dev]"

Quick start

vargate \
  --tumor  T1 \
  --normal N1 \
  --input  metrics/ \
  --output patient1_qc \
  --label  "Patient 1"

Produces patient1_qc.html (self-contained report) and patient1_qc.tsv.

Credits

• Logo and visual identity - Elisa Perrin
• Claude (Anthropic) - assisted with tests, documentation, refactoring, HTML report styling, script cleanup, and CLI extensions to cover more use cases

Citation

If you use VarGate in your work, please cite it via the CITATION.cff file or directly:

Natiez, P. (2026). VarGate (v1.0.0). Zenodo. https://doi.org/10.5281/zenodo.20376751

License

Distributed under the MIT License.