Skip to content

Environment Setup

Installation

Install STEER in a clean environment with Python 3.9 or newer. Some features also depend on R and mclust. The examples below show tested configurations rather than strict version locks.

Open Quick Start View Source

  • Core requirements

    Python 3.9+, PyTorch, PyG, torch-scatter, and torch-sparse.

  • Optional R support

    Some STEER features additionally require r-base and r-mclust.

  • Compatibility rule

    Match your CUDA, PyTorch, and PyG wheels carefully within the same environment.

Example GPU setup

Tested during development with a configuration such as:

  • GPU: NVIDIA 3090
  • CUDA: 12.4
  • PyTorch: 2.4.0
# Create and activate environment
conda create -n steer python=3.10
conda activate steer

# Optional R dependencies
conda install -c conda-forge r-base=4.3.3 r-mclust=6.1.1

# Install PyTorch with CUDA 12.4 support
pip install torch==2.4.0+cu124 torchvision==0.19.0 torchaudio==2.4.0 --index-url https://download.pytorch.org/whl/cu124

# Install PyG dependencies
pip install torch-scatter -f https://data.pyg.org/whl/torch-2.4.0+cu124.html
pip install torch-sparse -f https://data.pyg.org/whl/torch-2.4.0+cu124.html
pip install torch-geometric

# Install STEER
pip install git+https://github.com/lzygenomics/STEER.git

Tested newer setup

STEER has also been tested in a newer environment such as:

  • GPU: NVIDIA L20
  • CUDA: 12.8
  • PyTorch: 2.8.0
  • Python: 3.10
conda create -n steer python=3.10
conda activate steer

conda install -c conda-forge r-base=4.3.3 r-mclust=6.1.1

pip install torch==2.8.0 torchvision==0.23.0 torchaudio==2.8.0 --index-url https://download.pytorch.org/whl/cu128
pip install torch-scatter -f https://data.pyg.org/whl/torch-2.8.0+cu128.html
pip install torch-sparse -f https://data.pyg.org/whl/torch-2.8.0+cu128.html
pip install torch-geometric

# Install STEER
pip install git+https://github.com/lzygenomics/STEER.git

Notes

The environment versions listed above reflect configurations used during development and testing. They should be regarded as recommended examples rather than strict version requirements.

In general, STEER should work as long as the following packages are installed in a mutually compatible way:

  • torch
  • torch-geometric
  • torch-scatter
  • torch-sparse

Additional dependencies

For the broader Python environment, please also refer to the project requirements file. Key libraries include:

  • numpy
  • scipy
  • scanpy
  • scvelo
  • matplotlib
  • seaborn

See the repository requirements.txt for the full dependency list.

Troubleshooting

R dependency issues

If some STEER features fail because of missing R dependencies, make sure the following are available in your environment:

  • r-base
  • r-mclust

PyG installation issues

If torch-scatter or torch-sparse fail to install, check that:

  1. your CUDA version matches the installed PyTorch wheel
  2. the PyG wheel URL matches your PyTorch version
  3. you are installing into the intended virtual environment

Verify installation

After installation, open Python and test whether the main dependencies can be imported successfully.