DeepGFT: identifying spatial domains in spatial transcriptomics of complex and 3D tissue using deep learning and graph Fourier transform

System Requirments

OS requirements

DeepGFT can run on Linux and Windows. The package has been tested on the following systems:

• Linux: CentOS 7, NVIDIA Tesla V100 GPUs.
• Windows: Windows 10

DeepGFT requires python version >= 3.7. We tested it in python 3.8.16 and cuda 11.6.1 on Linux.

Installation Guide

Create a virtual environment

Users can install anaconda by following this tutorial if there is no Anaconda.

Create a separated virtual environment:

conda create -n DeepGFT python=3.8
conda activate DeepGFT

Install packages

Install r-base and mclust packages:

conda install -c conda-forge r=4.1.0
conda install -c conda-forge r-mclust

Install DeepGFT from Github and rpy2.

git clone https://github.com/jxLiu-bio/DeepGFT.git
cd DeepGFT
pip install -r requirement.txt
pip install rpy2==3.5.10

Next, run

python setup.py install

Install pytorch package of GPU version and pyG. See Pytorch and PyG and for detail. We passed the test on cuda 11.6.1. Users can choose the corresponding pytorch for other cuda versions. torch_sparse, torch_scatter, torch_cluster need to be manually downloaded on the pytorch-geometric.

pip install torch==1.12.0+cu116 torchvision==0.13.0+cu116 torchaudio==0.12.0 --extra-index-url https://download.pytorch.org/whl/cu116
pip install torch_sparse-0.6.16+pt112cu116-cp38-cp38-linux_x86_64.whl
pip install torch_scatter-2.1.0+pt112cu116-cp38-cp38-linux_x86_64.whl
pip install torch_cluster-1.6.0+pt112cu116-cp38-cp38-linux_x86_64.whl
pip install torch_geometric==2.1.0

Install jupyter notebook and set ipykernel.

conda install jupyter
python -m ipykernel install --user --name DeepGFT --display-name DeepGFT

Tutorial

For the step-by-step tutorial, please refer to: DeepGFT tutorial

Data availability

All data can be downloaded from https://drive.google.com/drive/folders/1uzrXJXbtwFomQuEldagfyA0Z_wfNqEza?usp=sharing.