Headless JSON API mode for external tool integration

[light-suzuki]

Problem

R/qtl2 is the gold standard for biparental QTL mapping, but calling it from external applications (Excel add-ins, Python wrappers, web services) requires either:

1. Writing intermediate CSV/TSV files and shelling out to Rscript
2. Using rpy2 which adds complex dependency management
3. Writing custom R wrapper scripts that parse CLI arguments

Each frontend re-invents this wrapper, leading to fragmentation and maintenance burden.

Feature request

An optional headless/JSON mode that:

• Accepts input via JSON (or JSON + CSV for large genotype matrices)
• Returns results as JSON (LOD scores, peaks, support intervals, permutation thresholds)
• Runs as a single CLI command
• Could be a thin wrapper package (e.g. churchill-lab/qtl2api)

Use case

ExcelMapBasedQTL (https://github.com/light-suzuki/Genotype) is an Excel VSTO add-in for map-based QTL narrowing of F2 populations. It currently integrates R/qtl2 via a custom r/qtl2_scan.R script. A standardized JSON API would make this integration reproducible across frontends and reduce wrapper code.

Other projects with the same need:

• QTL IciMapping headless pipeline integration
• Galaxy workflow system QTL tools
• Web-based QTL dashboards (Shiny/Streamlit)

Why this matters

• Reproducibility: a JSON contract is versionable and testable
• Interoperability: Python, C#, JavaScript frontends all speak JSON natively
• CI/validation: input/output schemas can be validated without R
• Lowers barrier: users of Excel/Galaxy/web tools can access R/qtl2 without writing R code

Would the maintainers consider adding or documenting a stable JSON-based headless interface?

[kbroman]

Thanks for the suggestion. It would be straightforward, once I've implemented Issue #12 (form a cross2 object direction from a set of data frames).