A selective JAK inhibitor screening and designing platform.
The code in this repository relies on the DGL package (https://github.com/dmlc/dgl) with pytorch backend (http://pytorch.org) as well as on sklearn (http://scikit-learn.org) and dgllife (https://github.com/awslabs/dgl-lifesci). We are recommended you to create a conda environment for example:
conda env create -f environment.yml
Then activate the environment:
conda activate MTATFP
First of all, we are required to prepare the molecules, taking 'C/C=C/C(O)=Nc1cccc(CNc2c(C(=N)O)cnn3cccc23)c1' as an exapmle:
python preprocess.py
And then, you can train the model, using the following command：
python MTATFP_train.py
Or, you could directly use our trained model with the preprocessed molecules to make predictions:
python MTATFP_test.py
The trained STATFP models were also provided here, you can also used them by the STATFP_test.py. And The LightGBM models for machine learning methods are also provided in repository.
You could open the atom_visualization.ipynb in Jupyter Notebooks to visualise atom features.