Skip to content

suinleelab/contrastiveVI

Repository files navigation

contrastiveVI

contrastiveVI is a generative model designed to isolate factors of variation specific to a group of "target" cells (e.g. from specimens with a given disease) from those shared with a group of "background" cells (e.g. from healthy specimens). contrastiveVI is implemented in scvi-tools.

User guide

Note: This implementation of contrastiveVI is no longer maintained. Please see the main scvi-tools repository for the most up to date version of contrastiveVI, along with an updated tutorial here.

Installation

To install the latest version of contrastiveVI via pip

pip install contrastive-vi

Installation should take no more than 5 minutes.

What you can do with contrastiveVI

  • If you have a dataset with cells in a background condition (e.g. from healthy controls) and a target condition (e.g. from diseased patients), you can train contrastiveVI to isolate latent factors of variation specific to the target cells from those shared with a background into separate latent spaces.
  • Run clustering algorithms on the target-specific latent space to discover sub-groups of target cells
  • Perform differential expression testing for discovered sub-groups of target cells using a procedure similar to that of scVI .

Colab Notebook Examples

Development guide

Set up the environment

  1. Git clone this repository.
  2. cd contrastive-vi.
  3. Create and activate the specified conda environment by running
    conda env create -f environment.yml
    conda activate contrastive-vi-env
    
  4. Install the constrative_vi package and necessary dependencies for development by running pip install -e ".[dev]".
  5. Git pre-commit hooks (https://pre-commit.com/) are used to automatically check and fix formatting errors before a Git commit happens. Run pre-commit install to install all the hooks.
  6. Test that the pre-commit hooks work by running pre-commit run --all-files.

Testing

It's a good practice to include unit tests during development. Run pytest tests to verify existing tests.

References

If you find contrastiveVI useful for your work, please consider citing our preprent:

@article{contrastiveVI,
  title={Isolating salient variations of interest in single-cell transcriptomic data with contrastiveVI},
  author={Weinberger, Ethan and Lin, Chris and Lee, Su-In},
  journal={bioRxiv},
  year={2021},
  publisher={Cold Spring Harbor Laboratory}
}

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages