Metadata-Version: 2.1
Name: scCausalVI
Version: 0.0.4
Summary: A deep causality-aware model for disentangling treatment effects at single-cell resolution for perturbational scRNA-seq data
Home-page: https://github.com/ShaokunAn/scCausalVI/
Author: Shaokun An
Author-email: shan12@bwh.harvard.edu
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.9
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: scanpy>=1.9.6
Requires-Dist: torch>=2.0.0
Requires-Dist: anndata>=0.10.3
Requires-Dist: numpy>=1.23.5
Requires-Dist: setuptools>=59.5.0
Requires-Dist: pandas>=2.1.1
Requires-Dist: matplotlib>=3.8.1
Requires-Dist: scikit-learn>=1.3.2
Requires-Dist: tqdm>=4.66.1
Requires-Dist: seaborn>=0.12.2
Requires-Dist: scipy>=1.11.3
Requires-Dist: scvi-tools>=0.16.1
Requires-Dist: pytorch-lightning>=1.5.10

# scCausalVI

[![Documentation Status](https://readthedocs.org/projects/scvi-tools/badge/?version=stable)](https://sccausalvi.readthedocs.io/en/latest/)
[![PyPI Downloads](https://static.pepy.tech/badge/sccausalvi)](https://pepy.tech/projects/sccausalvi)
[![PyPI Downloads](https://static.pepy.tech/badge/sccausalvi/month)](https://pepy.tech/projects/sccausalvi)

scCausalVI is a causality-aware generative model designed to disentangle inherent cellular heterogeneity from treatment effects in single-cell RNA sequencing data, particularly in case-control studies.

![scCausalVI Overview](https://github.com/ShaokunAn/scCausalVI/blob/main/sketch/method%20overview.png)


## Introduction

scCausalVI addresses a major analytical challenge in single-cell RNA sequencing: distinguishing inherent cellular variation from extrinsic cell-state-specific effects induced by external stimuli. The model:

- Decouples intrinsic cellular states from treatment effects through a deep structural causal network
- Explicitly models causal mechanisms governing cell-state-specific responses
- Enables cross-condition in silico prediction
- Accounts for technical variations in multi-source data integration
- Identifies treatment-responsive populations and molecular signatures

### Key Features of scCausalVI
- Interpretable and disentangled latent representation
- Data integration
- In silico perturbation
- Identification of treatment-responsive populations

## Installation
There are several alternative options to install scCausalVI:

1. Install the latest version of scCausalVI via pip:

   ```bash
   pip install scCausalVI
   ```

2. Or install the development version via pip:

   ```bash
   pip install git+https://github.com/ShaokunAn/scCausalVI.git
   ```

## Examples
See examples at our [documentation site](https://sccausalvi.readthedocs.io/en/latest/tutorial.html).

## Reproducing Results

In order to reproduce paper results visit [here](https://github.com/ShaokunAn/scCausalVI-reproducibility/tree/main).


## References

If you find this package useful, please cite:
TBD

## Contact

Feel free to contact us by [mail](shan12@bwh.harvard.edu). If you find a bug, please use the [issue tracker](https://github.com/ShaokunAn/scCausalVI/issues).
