MSF-DTA

Code for paper "Predicting drug-target affinity by learning protein knowledge from biological networks"

Dependencies

python == 3.7.11

pytorch == 1.7.1

PyG (torch-geometric) == 2.0.2

rdkit == 2020.09.5

numpy == 1.21.2

Data preparation (password: 1234)

Unzipping all ''.rar files'' to their path.
1. PPI network are available at Link.
2. Unzipping ''ppi.rar'' to ''MSF-DTA/data/networks/''.
3. Trained latent representations of 18,552 proteins are available at Link.
4. Unzipping ''embeddings.rar'' to ''MSF-DTA/data/''.
data splitting for Davis and KIBA.

python create_data.py
data splitting for Human

python create_data_for_CPI.py

DTA task (Davis and KIBA).

Evaluating trained model by us on Davis dataset.

python test.py 0 0 0
Evaluating trained model by us on KIBA dataset.

python test.py 1 0 0
If you want to train your own model on Davis dataset.

python training_validation.py.py 0 0 0
If you want to train your own model on KIBA dataset.

python training_validation.py.py 1 0 0

CPI task (Human).

Training and testing are combined.

python train_for_CPI.py

latent representations.

if you want to train your own latent representations.

python embeddings_gen.py

Cite our work

if you use the conclusion, code, or data in our work, please cite:

@ARTICLE{10027191,
  author={Ma, Wenjian and Zhang, Shugang and Li, Zhen and Jiang, Mingjian and Wang, Shuang and Guo, Nianfan and Li, Yuanfei and Bi, Xiangpeng and Jiang, Huasen and Wei, Zhiqiang},
  journal={IEEE Journal of Biomedical and Health Informatics}, 
  title={Predicting Drug-Target Affinity by Learning Protein Knowledge From Biological Networks}, 
  year={2023},
  volume={27},
  number={4},
  pages={2128-2137},
  keywords={Proteins;Protein engineering;Drugs;Amino acids;Feature extraction;Predictive models;Knowledge engineering;Drug-target affinity;variational graph auto-encoders;graph convolutional network;protein- 
  protein interaction},
  doi={10.1109/JBHI.2023.3240305}}

Wenjian-Ma/MSF-DTA