This is our PyTorch implementation for the paper:
GraphSynergy: Network Inspired Deep Learning Model for Anti-Cancer Drug Combination Prediction
GraphSynergy is a new deep learning framework to make explainable synergistic drug combination predictions. GraphSynergy is inspired by the recent network science studies on drug combination identifying task and utilizes Graph Convolutional Network (GCN) and attention module to capture the topological relations of the protein modules of drugs and cancer cell lines in the PPI network.
The code has been tested running under Python 3.7. The required package are as follows:
- pytorch == 1.6.0
- numpy == 1.19.1
- sklearn == 0.23.2
- networkx == 2.5
- pandas == 1.1.2
To install the required packages for running GraphSynergy, please use the following command first
pip install -r requirements.txt
If you meet any problems when installing pytorch, please refer to pytorch official website
- DrugCombDB
python train.py --config ./config/DrugCombDB_config.json
- Oncology-Screen
python train.py --config ./config/OncologyScreen_config.json
Datasets used in the paper:
- Protein-Protein Interaction Network is a comprehensive human interactome network.
- Drug-protein Associations are based on FDA-approved or clinically investigational drugs.
- Cell-protein Associations is harvested from the Cancer Cell Line Encyclopedia.
- DrugCombDB is a database with the largest number of drug combinations to date.
- Oncology-Screen is an unbiased oncology compound screen datasets.
Acknowledgement and thanks to others for open source work used in this project. Code used in this project is available from the following sources.
- https://github.com/victoresque/pytorch-template\
Author: SunQpark
Licensed under MIT License.