DSB_GNN

Requirements:
Python >= 3.6
PyTorch>=1.6
Numpy>=1.16.0
torch-geometric>=1.6.0
pandas>=0.24.0
scikit-learn>=0.20.0

GraphMaker

Save the NHEK datasets in the data directory, run the graphmaker.py file, and the running instructions are:
python Graphmaker.py
The generated file is automatically saved in data/NHEK_dgl_data folder.

Model training

The operation instructions are:
python NHEK_training.py --test_chr=X --device=device_num
Where, X represents the test chromosome number, ranging from 1 to 23, device_num is the GPU number. The trained model will be saved in the model folder.

Model test

The operation instructions are:
python NHEK_test.py --test_chr=X --device=device_num
The prediction results are saved in the prediction folder.

Model explain

The operation instructions are:
python NHEK_feature_explain.py --test_chr=X --device=device_num
python NHEK_edge_explain.py --test_chr=X --device=device_num
The feature importance and edge importance corresponding to the feature number will be generated in the current folder.