Installation

Requirements

• Python 3.12 or newer
• conda (recommended — it also provides the compression tools Arcane uses to read .gz/.xz FASTQ files)

Arcane is built on numba; most of its compute-intensive code is JIT-compiled at runtime. Dependencies are listed in environment.yml and pyproject.toml.

Install from source

Clone the repository:

git clone https://gitlab.com/rahmannlab/arcane
cd arcane

Create the conda environment from the provided environment.yml. This creates an environment named arcane with all required dependencies:

conda env create

Then activate it and install the package itself:

conda activate arcane
pip install -e .

Note

Run pip install -e . from the root of the cloned repository — the directory containing README.md and CHANGELOG.md.

Verify the installation:

arcane --version
arcane --help

This installs two console commands: arcane and scfastqsim (a small scRNA-seq FASTQ simulator used by the benchmarking workflows).

Prebuilt indices

Building an index for a full genome takes time and memory. For human and mouse we provide prebuilt indices:

:material-download: Download from Zenodo

These contain all gapped k-mers (k=31, w=43, mask ####_#_##_###_#_###_###_###_#_###_##_#_####) of all sequences from the CellRanger-filtered GTF annotation.

An index is a pair of files that share a prefix:

myindex.hash    # the hash table itself
myindex.info    # metadata (mask, k, rcmode, gene names, table parameters)

Throughout the docs, --index myindex refers to that shared prefix, and Arcane appends .hash and .info itself.

To build an index for a different species or a custom annotation, see Building an index.

Next steps

Head to the Quickstart.