Skip to content

This is the repository for the single-cell transcriptomics application of the Deep Generative Decoder (DGD), developed by the Krogh group.

License

Notifications You must be signed in to change notification settings

Center-for-Health-Data-Science/scDGD

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

24 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

scDGD

scDGD is an application of our encoder-less generative model, the Deep Generative Decoder (DGD), to single-cell transcriptomics data.

It learns low-dimensional representations of full transcriptomics matrices without feature selection. The low-dimensional embeddings are of higher quality than comparable methods such as scVI and the data reconstruction is highly data efficient, outperforming scVI and scVAE, especially on very small data sets.

For more information about the underlying method and our results, check out the paper.

Installation

You can install the package via

pip install git+https://github.com/Center-for-Health-Data-Science/scDGD

How to use it

From our experience, scDGD can be applied to data sets with as few as 500 cells and as many as one million.

Check out the notebook showing an example of how to use scDGD: Open In Colab

We have also uploaded pre-trained models for the PBMC (Zheng et al. 2017) and the 10X 1 million mouse brain data sets, along with notebooks showing how to load them in pretrained_models.

Reference

If you use scDGD in your research, please consider citing

@article{schuster_deep_2023,
	title = {The Deep Generative Decoder: MAP estimation of representations improves modelling of single-cell RNA data},
	volume = {39},
	issn = {1367-4811},
	shorttitle = {The Deep Generative Decoder},
	url = {https://doi.org/10.1093/bioinformatics/btad497},
	doi = {10.1093/bioinformatics/btad497},
	number = {9},
	journal = {Bioinformatics},
	author = {Schuster, Viktoria and Krogh, Anders},
	month = sep,
	year = {2023}
}

About

This is the repository for the single-cell transcriptomics application of the Deep Generative Decoder (DGD), developed by the Krogh group.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published