luslab / arshamg-scrnaseq-wgan

Wasserstein Generative Adversarial Network for analysing scRNAseq data

scRNAseq-WGAN-GP

Wasserstein Generative Adversarial Network for analysing scRNAseq data.

This repo contains a Jupyter Notebook with a minimal version of the WGAN-GP algorithm.

Related manuscript is on bioRxiv

Usage

Dependencies: Tensorflow, NumPy, Pandas, scikit-learn

Clone the repo, ensure you have git lfs installed and perform git lfs pull for the CSV files

python ./scripts/WGAN-GP_minimal.py

Watch discriminator/generator loss convergence:

tensorboard --logdir=./summaries/