CBIcall: a configuration-driven framework for variant calling in large sequencing cohorts

Variant calling for next-generation sequencing (NGS) data relies on a diverse ecosystem of tools and workflows. Large-scale collaborative studies increasingly adopt federated analysis, where each institution processes sensitive data locally using standardized pipelines. Deploying identical pipelines across multiple centers remains challenging because heterogeneous software environments and computing policies can cause workflow divergence and inconsistent results.

We developed CBIcall, a workflow backend-flexible, configuration-driven framework that runs standardized variant-calling pipelines from raw FASTQ files to analysis-ready VCFs. Users define each analysis in a single YAML parameters file, which CBIcall resolves against a controlled workflow registry and resource catalog. The execution driver validates parameters and checks compatibility among pipelines, analysis modes, workflow backends, genome builds, tool versions, and resource bundles. CBIcall supports reproducibility auditing by comparing executions using recorded provenance and output fingerprints. CBIcall dispatches validated workflows natively through Bash, Cromwell, Nextflow and Snakemake backends and provides production-ready pipelines for germline WES, WGS (single-sample or cohort joint genotyping following GATK Best Practices), and mitochondrial DNA analysis. We evaluated analytical performance using public benchmark datasets and validated reproducibility across four computing environments. We further deployed CBIcall in the EU HEREDITARY project, where it processed 1102 samples with both WES and mtDNA pipelines on an institutional HPC system, supporting its suitability for reproducible cohort-scale genomic analyses.

CBIcall is open source (GPLv3) and distributed with ready-to-run pipelines; full dependency and installation documentation is available at https://github.com/CNAG-Biomedical-Informatics/cbicall.