CGDock: Curvature-Aware Geometric Flow Framework for Protein-Ligand Docking

Overview

This repository contains the source code of CGDock. If you have questions, don't hesitate to open an issue or ask me via jialy5@mail2.sysu.edu.cn. I am happy to hear from you!

Setup Environment

conda create --name CGDock python=3.8
conda activate CGDock
conda install pytorch==1.12.0 torchvision==0.13.0 torchaudio==0.12.0 cudatoolkit=11.3 -c pytorch
pip install https://data.pyg.org/whl/torch-1.12.0%2Bcu113/torch_cluster-1.6.0%2Bpt112cu113-cp38-cp38-linux_x86_64.whl
pip install https://data.pyg.org/whl/torch-1.12.0%2Bcu113/torch_scatter-2.1.0%2Bpt112cu113-cp38-cp38-linux_x86_64.whl
pip install https://data.pyg.org/whl/torch-1.12.0%2Bcu113/torch_sparse-0.6.15%2Bpt112cu113-cp38-cp38-linux_x86_64.whl 
pip install https://data.pyg.org/whl/torch-1.12.0%2Bcu113/torch_spline_conv-1.2.1%2Bpt112cu113-cp38-cp38-linux_x86_64.whl
pip install https://data.pyg.org/whl/torch-1.12.0%2Bcu113/pyg_lib-0.2.0%2Bpt112cu113-cp38-cp38-linux_x86_64.whl
pip install torch-geometric==2.4.0 torchdrug==0.1.2 torchmetrics==0.10.2 tqdm mlcrate pyarrow accelerate Bio lmdb fair-esm tensorboard fair-esm rdkit-pypi==2021.03.4
conda install -c conda-forge openbabel 

Data

The PDBbind 2020 dataset can be download from http://www.pdbbind.org.cn. The training data we used comes from FAbind, which you can download via this link.

Before training or evaluation, please generate the ESM2 embedding vectors for the proteins based on the above preprocessed data.

python data_processing/generate_esm2_t33.py ${data_path}

Construct local curvature features and incorporate them into protein and ligand representations.

Training mode

Evaluate the model using the following code.

python test.py

Retrain the model using the following command.

accelerate launch main.py

About

Acknowledegments

We appreciate EquiBind, TankBind, E3Bind, DiffDock, FABind and other related works for their open-sourced contributions.