scGNN - single cell graph neural networks

https://raw.githubusercontent.com/hurraygong/scGNN/master/pictures/F1.large.jpg

scGNN (single cell graph neural networks) provides a hypothesis-free deep learning framework for scRNA-Seq analyses. This framework formulates and aggregates cell-cell relationships with graph neural networks and models heterogeneous gene expression patterns using a left-truncated mixture Gaussian model. scGNN integrates three iterative multi-modal autoencoders and outperforms existing tools for gene imputation and cell clustering on four benchmark scRNA-Seq datasets.

scGNN’s applications

  • Imputed gene expression matrix. to Models heterogeneous gene expression patterns using a left-truncated mixture Gaussian model.
  • Learned embedding (features) for clustering.
  • Learned graph edges of the cell graph in tuples: nodeA,nodeB,weights.
  • Identified cell types. Formulates and agregates cell-cell relationships with graph neural networks

Reference

Wang, J., Ma, A., Chang, Y. et al. scGNN is a novel graph neural network framework for single-cell RNA-Seq analyses. Nat Commun 12, 1882 (2021). https://doi.org/10.1038/s41467-021-22197-x

ACKNOWLEDGEMENTS

This work was supported by the National Institutes of Health’s National Institute of General Medical Sciences (awards R35-GM126985 and R01-GM131399).

Support

Feel free to submit an issue or send us an email. Your help to improve scGNN is highly appreciated.

Note

We recommend users to infer LTMG from their datasets. LTMG can improve performance on our benchmarks despite it consumes extra time in data preprocessing. We also provide supports without LTMG to save running time.