https://github.com/genentech/sVAE

https://github.com/genentech/sVAE

Last synced: 1 day ago
JSON representation

• Host: GitHub
• URL: https://github.com/genentech/sVAE
• Owner: Genentech
• License: apache-2.0
• Created: 2022-11-07T05:04:17.000Z (almost 2 years ago)
• Default Branch: main
• Last Pushed: 2023-07-21T14:07:07.000Z (about 1 year ago)
• Last Synced: 2024-01-03T12:34:17.114Z (9 months ago)
• Language: Python
• Size: 1.54 MB
• Stars: 49
• Watchers: 6
• Forks: 5
• Open Issues: 3
• Metadata Files:
  • Readme: README.md
  • License: LICENSE.md

Awesome Lists containing this project

• top-pharma50 - **genentech/sVAE** - 07-21 14:07:07 | (Ranked by starred repositories)

README

        
# Learning Causal Representations of Single Cells via Sparse Mechanism Shift Modeling

This repository contains an implementation of the sparse VAE framework applied to single-cell perturbation data, as descibed in ["Learning Causal Representations of Single Cells via Sparse Mechanism Shift Modeling"](https://arxiv.org/abs/2211.03553). 

[![Stars](https://img.shields.io/github/stars/Genentech/sVAE?logo=GitHub&color=yellow)](https://github.com/Genentech/sVAE/stargazers)

[![Code Style](https://img.shields.io/badge/code%20style-black-000000.svg)](https://github.com/python/black)

    

Overview of the sparse VAE framework applied to single-cell perturbation data. (A) Input data are gene expression profiles of cells under different genetic or chemical perturbations (colors), as well as the intervention label. (B) A schematic of the generative model, and the causal semantics of the sparse VAE (C) Three method outputs. (i) identification of target latent variables, encoded as a causal graph between the interventions and latent variables; (ii) a disentangled latent model for which individual latent variables are more likely to be interpreted as the activity of a relevant biological process; and (iii) the generalization of the generative model to unseen interventions (e.g., for latent target identification).

## User guide

### Installation

Download or clone this repository. Then from inside the folder simply run:

```

pip install -e . 

```

### Example

An example script for the sandbox can be found in ``` entry_points/demo.py```.

The code for reproducing the real data analysis can be found in ``` entry_points/run_real_data_replogle_wandb.py```.

## References

```

@article{svae+,

  title={Learning Causal Representations of Single Cells via Sparse Mechanism Shift Modeling},

  author={Lopez, Romain and Tagasovska, Natasa and Ra, Stephen and Cho, Kyunghyun and Pritchard, Jonathan K. and Regev, Aviv },

  journal={Conference on Causal Learning and Reasoning},

  year={2023},

}

```