Learning Subpocket Prototypes for Generalizable Structure-based Drug Design

Official Pytorch implementation of ICML'23 paper "Learning Subpocket Prototypes for Generalizable Structure-based Drug Design" (https://arxiv.org/abs/2305.13997).

Previous work

• DrugGPS is based on our previous work on structure-based drug design: Molecule Generation For Target Protein Binding With Structural Motifs (ICLR 2023)
  • Code: https://github.com/zaixizhang/FLAG
  • Paper: https://openreview.net/forum?id=Rq13idF0F73

Install conda environment via conda yaml file

conda env create -f druggps_env.yaml
conda activate druggps_env

Datasets

Please refer to README.md in the data folder.

Preprocess dataset with BINANA

cd utils
python preprocess_data.py

Training

python train.py

Sampling

python sample.py

Generated Molecules for CrossDocked dataset

The generated molecules for 100 protein targets are stored in druggps.pt

The index file is test_index.pkl

Reference

@article{zhang2023learning,
  title={Learning Subpocket Prototypes for Generalizable Structure-based Drug Design},
  author={Zhang, Zaixi and Liu, Qi},
  journal={ICML},
  year={2023}
}