An inductive graph neural network model for compound-protein interaction prediction based on a homogeneous graph.

Installation with conda

If you don't have conda installed, please install it following the instructions here.

git clone https://github.com/wanxiaozhe/CPI-IGAE

cd CPI-IGAE

conda env create -f environment.yml

Datasets

All datasets are stored in ~/dataset

Usage

To check the results of test set, please run:

$ python test.py

To check the results of two external datasets DrugBank and TTD, please run:

$ python outtest.py --outtest xxx

xxx can be drugbank or ttd

optional arguments:
  -h, --help            show this help message and exit
  --batch_size	        default is 512
  --device              which cuda device to use (-1 for cpu training)
                            default is 0, cpu is not recommended due to to cpu 
                            due to the long runing time
  --model               default is the trained model in `~/best_model`
  --outtest             only in the `outtest.py`, can be `drugbank` or `ttd`

Our trained model is ~/best_model/final_model.pth