/drug-interactions

Project of drug interactions https://arxiv.org/pdf/2209.09941.pdf https://arxiv.org/pdf/2302.08680.pdf, NeurIPS 2022 (AI for Science)

Primary LanguagePythonMIT LicenseMIT

Project of drug interactions

Contributors:

  • Ngo Nhat Khang
  • Hy Truong Son (Correspondent / PI)

Papers:

Decagon

Figure taken from http://snap.stanford.edu/decagon/

Requirements

Data

Make sure a Data folder is created in each data's subfolder. Then, you should donwload data from the links below and locate them into the Data folders as:

  ├── Anticancer                
  │   ├── Data
  │   ├── ...
  └── Polypharmacy                
  │   ├── Data  
  │   ├── ...
  └── README.md

Anticancer

Download from ADRML

Polypharmacy

Download from Decagon

Run

Anticancer

  cd Anticancer/
  bash train.sh

Polypharmacy

  • Train GAE
      cd Polypharmacy/
      python3 train_hetero_gae.py --seed 1 --num_epoch 300 --lr 1e-3 --chkpt_dir ./ --dropout 0.1 --device cuda:0
  • Train VGAE
      cd Polypharmacy/
      python3 train_hetero_vgae.py --seed 1 --num_epoch 300 --lr 1e-3 --chkpt_dir ./ --dropout 0.1 --device cuda:0 --latent_encoder_type linear
  • Train VGAE + Morgan fingerprints
      cd Polypharmacy/
      python3 train_hetero_vgae_morgan.py --seed 1 --num_epoch 300 --lr 1e-3 --chkpt_dir ./ --dropout 0.1 --device cuda:0 --latent_encoder_type linear

Citations

@article{Zitnik2018,
  title={Modeling polypharmacy side effects with graph convolutional networks},
  author={Zitnik, Marinka and Agrawal, Monica and Leskovec, Jure},
  journal={Bioinformatics},
  volume={34},
  number={13},
  pages={457–466},
  year={2018}
}
@article{ahmadi2020adrml,
  title={ADRML: anticancer drug response prediction using manifold learning},
  author={Ahmadi Moughari, Fatemeh and Eslahchi, Changiz},
  journal={Scientific reports},
  volume={10},
  number={1},
  pages={1--18},
  year={2020},
  publisher={Nature Publishing Group}
}