Deep Reinforcement Learning for de-novo Drug Design
This is an official PyTorch implementation of Deep Reinforcement Learning for de-novo Drug Design aka ReLeaSE method.
In order to get started you will need:
If you installed your Python with Anacoda you can run the following commands to get started:
# Clone the reopsitory to your desired directory
git clone https://github.com/isayev/ReLeaSE.git
cd ReLeaSE
# Create new conda environment with Python 3.6
conda create -n release python=3.6
# Activate the environment
conda activate release
# Install conda dependencies
conda install --yes --file conda_requirements.txt
conda install -c rdkit rdkit nox cairo
conda install pytorch=0.4.1 torchvision=0.2.1 -c pytorch
# Instal pip dependencies
pip install pip_requirements.txt
# Add new kernel to the list of jupyter notebook kernels
python -m ipykernel install --user --name release --display-name ReLeaSE
We uploaded several demos in a form of iPython notebooks:
Disclaimer: JAK2 demo uses Random Forest predictor instead of Recurrent Neural Network, since the availability of the dataset with JAK2 activity data used in the "Deep Reinforcement Learning for de novo Drug Design" paper is restricted under the license agreement. So instead we use the JAK2 activity data downladed from ChEMBL (CHEMBL2971) and curated. The size of this dataset is ~2000 data points, which is not enough to build a reliable deep neural network. If you want to see a demo with RNN, please checkout logP optimization.
If you use this code or data, please cite:
Mariya Popova, Olexandr Isayev, Alexander Tropsha. Deep Reinforcement Learning for de-novo Drug Design. Science Advances, 2018, Vol. 4, no. 7, eaap7885. DOI: 10.1126/sciadv.aap7885