Installation

Contents

Installation#

bio2zarr can either be installed from bioconda or PyPI.

By default the bio2zarr PyPI package does not come with dependencies needed to convert VCF, BGEN or tskit files, so optional dependencies need to be specified:

python3 -m pip install bio2zarr #Base package with only plink support
python3 -m pip install bio2zarr[vcf] #Install with VCF support
python3 -m pip install bio2zarr[tskit] #Install with tskit support
python3 -m pip install 'bio2zarr[bgen]' #Install with BGEN support
python3 -m pip install bio2zarr[all] #Install with all optional dependencies

For bioconda users, a package is available in the bioconda channel:

conda install -c bioconda bio2zarr

This will install the programs vcf2zarr, vcf_partition, tskit2zarr bgen2zarr and plink2zarr into your local Python path. You may need to update your $PATH to call the executables directly.

BGEN support uses the bgen-limix distribution, version 4.15.0 or later (below 5), from jpfeuffer/bgen-limix. Its import namespace is named cbgen, but it is not the legacy cbgen distribution. Do not install the two distributions in the same environment, because their import namespaces conflict. Python 3.11+ and NumPy 2+ are required. Backend wheels target macOS ARM64, Linux x86-64/ARM64 and Windows x86-64; other platforms, including Intel macOS, require a source build with the backend’s native build dependencies (C/C++, CMake, zlib and zstd). Bioconda packaging may lag PyPI’s BGEN extra.

Alternatively, calling for example:

python3 -m bio2zarr vcf2zarr <args>

is equivalent to

vcf2zarr <args>

and will always work.

Warning

Windows support is preliminary, partial and needs to be fully documented. We recommend trying the bioconda packages in the first instance, and if this doesn’t work try using Windows Subsystem for Linux (WSL). Please comment on this issue if you would like to see improved Windows support for bio2zarr, or would like to help out with this.

Shell completion#

To enable shell completion for a particular session in Bash do:

eval "$(_VCF2ZARR_COMPLETE=bash_source vcf2zarr)"
eval "$(_TSKIT2ZARR_COMPLETE=bash_source tskit2zarr)"
eval "$(_PLINK2ZARR_COMPLETE=bash_source plink2zarr)"
eval "$(_BGEN2ZARR_COMPLETE=bash_source bgen2zarr)"

If you add this to your .bashrc shell completion should available in all new shell sessions.

See the Click documentation for instructions on how to enable completion in other shells.

Contents