/DeepDTAF

a deep learning architecture for protein-ligand binding affinity prediction

Primary LanguageJupyter NotebookApache License 2.0Apache-2.0

About DeepDTAF

DeepDTAF is a deep learning architecture, which integrates local and global features to predict the binding affinity between ligands and proteins.

The benchmark dataset can be found in ./data/. Data preprocessing can be referred to ./prepare/. The DeepDTAF model is available in ./src/. And the result will be generated in ./runs/. See our paper for more details.

Requirements:

  • python 3.7
  • cudatoolkit 10.1.243
  • cudnn 7.6.0
  • pytorch 1.4.0
  • numpy 1.16.4
  • scikit-learn 0.21.2
  • pandas 0.24.2
  • tensorboard 2.0.0
  • scipy 1.3.0
  • numba 0.44.1
  • tqdm 4.32.1

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 DeepDTAF_env

Then, install the apex in the DeepDTAF_env environment:

git clone https://github.com/NVIDIA/apex
cd apex
python setup.py install

Since the codes for apex package in the above website have been updated, you can also install it using the uploaded package. The apex.tar can be found in ./src/.

Training & Evaluation

to train your own model

cd ./src/
python main.py

to see the result

tensorboard ../runs/DeepDTAF_<datetime>_<seed>/

Citation

Wang K, Zhou R, Li Y, et al. DeepDTAF: a deep learning method to predict protein–ligand binding affinity. Briefings in Bioinformatics, 2021, 22(5), bbab072.

contact

Kaili Wang: kailiwang@dhu.edu.cn