This is the repository for our Bioinformatics article submission "MGOL: Molecule Generation with Ordering Loss" by Vincent HUANG, Selim YAHIA-MESSAOUD, Wenming YANG, Dong-Qing WEI, Jie CHEN , Yonghong TIAN.
This code is adapted from the official repository of the 2022 ICML paper "LIMO: Latent Inceptionism for Targeted Molecule Generation" by Peter Eckmann, Kunyang Sun, Bo Zhao, Mudong Feng, Michael K. Gilson, and Rose Yu.
Please ensure that RDKit and Open Babel are installed. The following Python packages are also required (these can also be installed with pip install -r requirements.txt):
torch
pytorch-lightning
selfies
scipy
tqdm
To optimize molecules for binding affinity, an AutoDock-GPU executable must be compiled. To generate your own protein files, see Steps 1-2 in the AutoDock4 User Guide. The compiled file has to be renamed 'autodock_gpu_128wi' and put into the main directory. The AutoDock-GPU Wiki may also be helpful.
To run the code that leads to the results presented in the Table 3, you should run :
''' bash run.sh '''