Installation¶
We tested TFcomb with the Python environment 3.9.
conda create --name TFcomb python=3.9
Prerequisites¶
Before installing TFcomb, users should first install Pytorch and DGL.
Check your CUDA version and install Pytorch with the right version. For example, the CUDA version of our server is 11.3, so we refered the website and install Pytorch with:
conda install pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cudatoolkit=11.3 -c pytorch
The versions of DGL can be refered at the website. According to our CUDA version, we installed DGL with:
conda install dglteam/label/cu113::dgl
By the way, if the install speed of conda is too slow, you can just install mamba and replace conda with mamba:
conda install mamba
mamba install dglteam/label/cu113::dgl
After Pytorch and DGL are installed in the environment, you can install TFcomb through pip or github.
PyPI¶
TFcomb is available on PyPI here and can be installed via:
pip install bio-tfcomb
GitHub¶
PAST can also installed from GitHub via:
git clone git@github.com:Chen-Li-17/TFcomb.git
cd TFcomb
python setup.py install
Dependency of test env¶
adjustText==1.3.0
auto_mix_prep==0.2.0
celloracle==0.12.0
dgl==1.1.1
gseapy==1.1.4
matplotlib==3.6.3
networkx==3.2.1
numpy==1.26.4
pandas==1.5.3
scanpy==1.10.3
scikit_learn==1.2.2
scipy==1.13.1
seaborn==0.13.2
torch==1.12.1
tqdm==4.67.1
umap_learn==0.5.7
If there are any problems with your installation, you can refer to the specific versions of the packages.