Installation¶
It’s recommended to create a new virtual environment before installing the dependencies:
conda create -n STADiffuser python=3.9
Activate the virtual environment:
conda activate STADiffuser
Prerequisites¶
Make sure you have installed a version of PyTorch (version >= 1.13.1) that is compatible with your GPU (if applicable) first.
pip install torch==1.13.1+cu117 torchvision==0.14.1+cu117 torchaudio==0.13.1 --extra-index-url https://download.pytorch.org/whl/cu117
Replace cu117 with the appropriate CUDA version for your system if it differs from CUDA 11.8. If you are using a CPU-only version, you can simply install PyTorch without specifying a CUDA version.
Install torch_geometric dependencies:
pip install torch_scatter torch_sparse torch_cluster torch_spline_conv -f https://data.pyg.org/whl/torch-1.13.1+cu117.html
Install from source code¶
To install STADiffuser from source code, clone the repository and install the dependencies using the following commands:
git clone git@github.com:messcode/STADiffuser.git
Now you can install the code from the source directory:
cd STADiffuser
pip install -e .
Install from PyPI¶
To install STADiffuser from PyPI, use the following command:
pip install stadiffuser
Trouble shooting¶
If you fail to install, you may need to install the following dependencies manually:
SCANPY: recommended version 1.9.3
PyTorch: recommended version 1.13.1
PyG (PyTorch Geometric): recommended version 2.3.1