Skip to main content

Installation

CellType-Refinery can be installed from source using pip.

Requirements

  • Python >= 3.10
  • pip

From Source

Clone the repository and install in editable mode:

git clone https://github.com/kimpenn/CellType-Refinery.git
cd CellType-Refinery
pip install -e .

Optional Dependencies

Visualization Support

For generating plots and figures:

pip install -e ".[viz]"

This includes matplotlib and seaborn.

GPU Acceleration

For RAPIDS/cuGraph-accelerated Leiden clustering:

pip install -e ".[gpu]"

This includes rapids-singlecell and cupy-cuda12x.

Development Tools

For testing and code quality:

pip install -e ".[dev]"

This includes pytest, pytest-cov, ruff, and mypy.

All Dependencies

Install everything:

pip install -e ".[all]"

Verify Installation

After installation, verify CellType-Refinery is working:

# Check CLI is available
celltype-refinery --help

# Check Python import
python -c "from celltype_refinery.core.annotation import AnnotationEngine; print('OK')"

Dependencies

Core Dependencies

PackageVersionPurpose
numpy>= 1.24Numerical computing
pandas>= 2.0Data manipulation
scipy>= 1.10Statistical functions
scikit-learn>= 1.2Machine learning utilities
anndata>= 0.10AnnData objects
scanpy>= 1.9Single-cell analysis
leidenalg>= 0.10Leiden clustering
pyyaml>= 6.0Configuration files
click>= 8.0CLI framework
tqdm>= 4.65Progress bars
statsmodels>= 0.14Statistical models

Optional Dependencies

ExtraPackagesPurpose
vizmatplotlib, seabornVisualization
gpurapids-singlecell, cupyGPU acceleration
devpytest, ruff, mypyDevelopment tools