Multiscale differential geometry learning of networks with applications to single-cell RNA sequencing data

This script is for the paper "Multiscale differential geometry learning of networks with applications to single-cell RNA sequencing data, Hongsong Feng, Sean Cottrell, Yuta Hozumi, and Guo-Wei Wei".

Requirements

Python Dependencies

python (>=3.7)
numpy (1.17.4)
scikit-learn (0.23.2)
scipy (1.5.2)
pandas (0.25.3)

Download the repository and scRNA-seq data

Download the repository from Github

# download repository by git
git clone https://github.com/WeilabMSU/MDG.git

Download the scRNA-seq data and CCP-UMAP features under the downloaded MDG folder.

cd MDG
wget https://weilab.math.msu.edu/Downloads/MDG/features-CCP-UMAP.zip  
wget https://weilab.math.msu.edu/Downloads/MDG/scRNA-seq-data.zip  
unzip features-CCP-UMAP.zip  
unzip scRNA-seq-data.zip

Generating differential geometry features for MDG modeling.

# use dataset GSE45719 for demonstration and kappa is set to 5 and 10 in our paper.
cd MDG
python mdg-curvature.py --dataset_name GSE45719 --kappa 5

The generated features are saved in the folder "features-CCP-UMAP/{dataset}_features".

Build MDG models and carry out five-fold cross-validations.

cd MDG
python MDG-classification.py --dataset_name GSE45719

Reference

Hongsong Feng, Sean Cottrell, Yuta Hozumi, and Guo-Wei Wei, "Multiscale differential geometry learning of networks with applications to single-cell RNA sequencing data" , Computers in Biology and Medicine, 171(2024): 108211.

License

All codes released in this study is under the MIT License.