PerturbNet is a deep generative model that can predict the distribution of cell states induced by chemical or genetic perturbation. The repository contains the data and model for reproducing the results. Currently, you can refer to the preprint PerturbNet predicts single-cell responses to unseen chemical and genetic perturbations. We will submit an updated version of the paper soon.

The current version of PerturbNet requires Python 3.7. All required dependencies are listed in requirements.txt in our github. We recommend creating a clean Conda environment using the following command:

After setting up the environment, you can install the package by running:

The tutorials and latest code are available on our GitHub. PerturbNet

data_paper contains all the data necessary for reproducing the benchmarks and results in the paper.

example_data contains all toy datasets for running the tutorials.

models contains model weights for reproducing the results in the paper, including weights for PerturbNet, chemCPA, GEARS, and Biolord.

pretrained_model contains the model weights of pretrained genotypeVAE and chemicalVAE. Data for generating the one-hot matrix for genetic perturbations is also included.