welch-lab / MichiGAN

Learning disentangled representations of single-cell data for high-quality generation

MichiGAN: sampling from disentangled representations of single-cell data using generative adversarial networks

Sampling from disentangled representations of single-cell data using generative adversarial networks

The current folder contains files for implementing PCA, GMM, VAE/beta-TCVAE, WGAN-GP, InfoWGAN-GP, ssInfoWGAN-GP, CWGAN-GP and MichiGAN on single-cell RNA-seq data. See our preprint for details:
Sampling from disentangled representations of single-cell data using generative adversarial networks (Yu and Welch, 2021+). We have a presentation video for Learning Meaningful Representations of Life Workshop at NeurIPS 2020, where we referred to our framework as DRGAN. We decided to change the name to MichiGAN afterwards.

List of Files:

1. /data is the folder containing the real scRNA-seq dataset of Tabula Muris heart data. Users can download the SCANPY-processed data on https://www.dropbox.com/sh/xseb0u6p01te3vr/AACuskVfswUFn5MroEFrqI-Xa?dl=0.

2. /FunctionProgramExamples/examples is the folder for the experiments of
   (1) vae.py: VAE;
   (2) beta_tcvae.py: beta-TCVAE;
   (3) wgangp.py: WGAN-GP;
   (4) infowgangp.py: InfoWGAN-GP;
   (5) MichiGAN_mean.py: MichiGAN on mean representations;
   (6) MichiGAN_sample.py: MichiGAN on sampled representations
   on the Tabula Muris heart dataset. The ipython notebooks example_**.ipynb also give examples of how to train different deep generative models on Tabula Muris heart data.

3. /FunctionProgramExamples/Adam_prediction.py is the StableGAN implementation file on https://github.com/taki0112/StableGAN-Tensorflow for the GAN-based methods

4. /FunctionProgramExamples/lib.py contains the Python and TensorFlow functions

5. /FunctionProgramExamples/nets.py has the network architectures for different deep generative models

Notes:

1. The example program demonstrates the training for Tabula Muris data with 4221 cells and 4062 genes processed by the SCANPY package and stored as .npy file.
2. The related module versions are:

 (1) Python 3.6
 (2) numpy: 1.16.3
 (3) pandas 0.25.3 
 (4) scanpy: 1.4.6
 (5) tensorflow: 1.14.0 

Reference

Please consider citing

@article{yu2021michigan,
  title={MichiGAN: sampling from disentangled representations of single-cell data using generative adversarial networks},
  author={Yu, Hengshi and Welch, Joshua D},
  journal={Genome biology},
  volume={22},
  number={1},
  pages={1--26},
  year={2021},
  publisher={BioMed Central}
}

We appreciate your interest in our work.