GraphCpG

Dataset	Cell Num	(AUROC) CaMelia	(AUROC) DeepCpG	(AUROC) CpG Transformer	(AUROC) GraphCpG
HCC	25	97.11	96.01	97.56	96.97
MBL	30	89.36	87.12	92.05	89.73
Hemato	122	87.68	88.26	89.56	89.77
Neuron-Mouse	690	91.13	88.59	90.87	91.75
Neuron-Homo	780	92.98	90.06	92.31	93.2

Visualization of locus-aware neighboring subgraphs from HCC

Image text

Files

Model
- graphcpg
- training
- imputation
- visualization
Data
- Neuron-Mouse
- Neuron-Homo
Analysis
- hierarchical analysis
- differential methylation analysis

Environment and Packages

python 3.9
cuda 11.1
pytorch 1.9.1+cu111
torhcvision 0.10.1+cu111
torchaudio 0.9.1
pytorch_geometric
torch-sparse
torch-geometric
pytorch-lightning 1.7.7

if error (module 'distutils'), please install setuptools 59.5.0

requirement.txt is also provided for reference, but compatibility still depends on the devices.

pytorch-lightning version incompatibility may cause extremely long training time and degraded result. (millions of thanks for colleagues asking questions related to this problem by emailing and in issues)

Usage

Enter the model folder

python train_graph_cpg.py

Citation

GraphCpG: Imputation of Single-cell Methylomes Based on Locus-aware Neighboring Subgraphs

@article{Deng2023GraphCpGIO,
  title={GraphCpG: imputation of single-cell methylomes based on locus-aware neighboring subgraphs},
  author={Yuzhong Deng and Jianxiong Tang and Jiyang Zhang and Jianxiao Zou and Que Zhu and Shicai Fan},
  journal={Bioinformatics},
  year={2023},
  volume={39},
  url={https://api.semanticscholar.org/CorpusID:261381228}
}

License

MIT license

yuzhong-deng / graphcpg

readme