IPDiff-ICLR 2024

Official implementation for our ICLR 2024 paper Protein-Ligand Interaction Prior for Binding-aware 3D Molecule Diffusion Models.

Environment

conda env create -f ipdiff.yaml
conda activate ipdiff

Data and Preparation

The data preparation follows TargetDiff. For more details, please refer to the repository of TargetDiff.

❗️❗️Path to Pretrained IPNet:

./pretrained_models

❗️❗️Pretrained IPDiff:

Download the pretrained model checkpoint of IPDiff from pretrained-ipdiff and move it to this path:

./pretrained_models

Training

conda activate ipdiff
python train.py

Sampling

python sample_split.py --start_index 0 --end_index 99 --batch_size 25

Evaluation

python eval_split.py --eval_start_index 0 --eval_end_index 99

Calculate metrics

python cal_metrics_from_pt.py

Citation

@inproceedings{huang2024proteinligand,
  title={Protein-Ligand Interaction Prior for Binding-aware 3D Molecule Diffusion Models},
  author={Zhilin Huang and Ling Yang and Xiangxin Zhou and Zhilong Zhang and Wentao Zhang and Xiawu Zheng and Jie Chen and Yu Wang and Bin CUI and Wenming Yang},
  booktitle={International Conference on Learning Representations},
  year={2024}
}