ACARL

This is the code repository for our paper published in Journal of Cheminformatics: Activity cliff-aware reinforcement learning for de novo drug design.

Dependencies:

pytorch
numpy
pandas
tqdm
tensorboard
rdkit
openbabel
PyTDC

Run the ACARL algorithm (for example, design drug molecules against the 5HT1B target):

python codes/pretrain.py
python codes/RLtrain.py --target 5HT1B