Non-containerized installation

Install the CBIcall Python package first. The large reference and software bundle is separate and is required only for the bundled cbicall-core workflows.

1. Install CBIcall

PyPI is recommended for normal use:

python3 -m pip install --upgrade cbicall
cbicall --version

Install optional Snakemake and MultiQC dependencies when needed:

python3 -m pip install --upgrade "cbicall[all]"

Nextflow and Cromwell are external executables and are not installed as Python dependencies. Bash workflows do not require either engine.

2. Choose the workflow source

Workflow source	CBIcall bundle required?	Additional runtime
workflow_provider: cbicall-core (default)	Yes	The selected backend: Bash, Snakemake, Nextflow, or Cromwell. mtDNA also requires x86_64 Linux.
workflow_provider: nf-core	No	Nextflow and a working nf-core runtime profile, such as Docker or Apptainer.

To test external nf-core support without installing the CBIcall bundle:

cbicall test --nf-core-demo -t 4

3. Install resources for cbicall-core

Choose a persistent directory and install the registered bundle:

export CBICALL_DATA=/absolute/path/to/cbicall-data
mkdir -p "$CBICALL_DATA"
cbicall install-resources --outdir "$CBICALL_DATA"

The installer downloads and assembles the bundle, verifies its catalog-declared checksums, extracts Databases/ and NGSutils/, and writes cbicall-resource-installation.json with installation provenance.

Set CBICALL_DATA in every shell or scheduler environment used to run bundled workflows:

export CBICALL_DATA=/absolute/path/to/cbicall-data
cbicall doctor

The same environment variable is propagated to all four bundled WES/WGS backends and recorded in the execution contract.

4. Run an integration test

cbicall test --wes-bash -t 1

On x86_64, the mtDNA workflow can also be tested with:

cbicall test --mit-bash -t 1

Installed-package tests use packaged fixtures and create a temporary workspace. Use --workspace /path/to/new-or-empty-directory to retain the test outputs in a chosen location.

Source checkout for development

Clone the repository and install the minimum functional Python package:

git clone https://github.com/CNAG-Biomedical-Informatics/cbicall.git
cd cbicall
python3 -m pip install -e .
cbicall --version

This installs CBIcall and its core dependencies only. For a complete development environment with Snakemake, MultiQC, and the test tools, use:

python3 -m pip install -e ".[all,test]"
pytest

Choose the extras you need

Editable install	What it adds
-e .	Core CBIcall dependencies, including PyYAML, JSON Schema validation, and resource downloading.
-e ".[snakemake]"	Snakemake and its pinned PuLP dependency.
-e ".[multiqc]"	MultiQC report generation.
-e ".[all]"	Both Snakemake and MultiQC.
-e ".[test]"	pytest and coverage tooling.
-e ".[all,test]"	All optional runtime integrations plus the test suite; this is the complete development installation shown above.
-e ".[build]"	Package build and upload tools used for releases.

Nextflow and Cromwell are not Python extras and must be installed separately.

Use cbicall after installation. The repository launcher bin/cbicall remains available inside a source checkout.

Slurm

For a source checkout on the CNAG GenE cluster, see the complete run_cbicall_slurm.sh template. It selects the appropriate research partition, requests the workflow resources, exposes the Python packages installed under the site-specific cbi_py3 prefix, and launches CBIcall with --runtime-profile cnag-hpc.

Manual resource-download recovery

Google Drive can restrict large automated downloads. Print the registered file list when automatic download fails:

cbicall install-resources \
  --outdir "$CBICALL_DATA" \
  --print-manual-download

Place every listed file in $CBICALL_DATA, then resume verification, assembly, and extraction:

cbicall install-resources \
  --outdir "$CBICALL_DATA" \
  --skip-download

Use --verify-resource-id-only to verify the small catalog-pinned resource identifier before downloading the archive. Add --remove-parts after successful verification when disk space is limited.

System requirements

• Linux on amd64 or arm64; macOS can be used through a Linux VM.
• Python 3.8 or newer.
• Java 17 for GATK 4.6 workflows.
• libncurses.so.5 and libtinfo.so.5 for bundled legacy tools.
• Sufficient storage for the resource bundle and analysis inputs; allow at least 100 GB for installation and the shipped WES test.

The bundled MToolBox mtDNA workflow supports x86_64 only.