/GraphscoreDTA

A novel graph neural network strategy with the Vina distance optimization terms to predict protein-ligand binding affinity

Primary LanguagePythonGNU General Public License v3.0GPL-3.0

About GraphscoreDTA

GraphscoreDTA is an optimized graph neural network for protein-ligand binding affinity prediction.

The benchmark dataset can be found in ./test_set/. The GraphscoreDTA model is available in ./src/. And the result will be generated in ./result/. See our paper for more details.

[IMPORTANT] We provide the input file in the release page. Please download it to ./test_set/.

Requirements:

  • python 3.7.11
  • pytorch 1.9.0
  • scikit-learn 0.24.2
  • dgl 0.9.1.post1
  • tqdm 4.62.2
  • ipython 7.27.0
  • numpy 1.20.3
  • pandas 1.3.2
  • numba 0.53.1
  • scipy 1.7.1

Installation

In order to get GraphscoreDTA, you need to clone this repo:

git clone https://github.com/CSUBioGroup/GraphscoreDTA
cd GraphscoreDTA

The easiest way to install the required packages is to create environment with GPU-enabled version:

conda env create -f environment_gpu.yml
conda activate GraphscoreDTA

Predict

to use our model

cd ./src/
python predict.py

Training

to train your own model

cd ./src/
python train.py

Citation

Wang K, Zhou R, Tang J, et al. GraphscoreDTA: optimized graph neural network for protein–ligand binding affinity prediction[J]. Bioinformatics, 2023, 39(6): btad340.

Contact

Kaili Wang: kailiwang@dhu.edu.cn

You can also download the codes from https://github.com/KailiWang1/GraphscoreDTA