This is the accompanying code for the paper, "Multi-Domain Translation between Single-Cell Imaging and Sequencing Data using Autoencoders" (bioRxiv)

The preprocessing scripts for the raw data files can be found at this repo (link).

The preprocessed data files can be downloaded from Dropbox (link).

Packages are listed in environment.yml file and can be installed using Anaconda/Miniconda:

conda env create -f environment.yml
conda activate pytorch

This code was tested on NVIDIA GTX 1080TI GPU.

Training the image autoencoder with classifier in latent space:

python train_ae.py --save-dir <path/to/save/dir> --conditional

Integrating the RNA autoencoder with conditional discriminator in latent space:

python train_rna_image.py --save-dir <path/to/save/dir> --pretrained-file <path/to/pretrained/image/autoencoder> --conditional-adv

Output is log file and PyTorch checkpoint files when code is run on gene expression and imaging data.