Dynamic graph attention network for predicting drug-target binding affinity. It is built with *Pytorch and Python 3.
- python 3.9, numpy, scipy, pandas, pytorch, pyg
# create
conda create -n DGDTA python=3.9
# activate
conda activate DGDTA
# deactivate
conda deactivate
- After creating and activating the DGDTA virtual environment, download DGDTA from github:
git clone https://github.com/luojunwei/DGDTA.git
cd DGDTA
conda activate DGDTA
conda install numpy, scipy, pandas, Pytorch, pyg
The example data can be downloaded from
https://github.com/thinng/GraphDTA/tree/master/data
python data_creation.py
First, divide the data into training and test sets and create data files in pytorch format.
Run the following script to train the model.
python training2.py
The default values of the parameter parser are the DGDTA-CL version and the KIBA dataset.
Run the following script to test the model.
python training_validation.py
This returns the best MSE model for the validation dataset during the training process.