This code implements the following paper:

Interpretable Sparsification of Brain Graphs: Better Practices and Effective Designs for Graph Neural Networks

• PyTorch
• PyTorch-Geometric
• numpy

To train our model, run the following script:

python ./src/main.py --method IGS --label_col $TASK_NAME --model $MODEL_NAME

where --label_col specifies the name of the task you want to work on and --model specifies the backbone model you want to use.

We have provided the data in ./data directory. To generate your own random data splits, you could use process_data_splits from ./data/data_splits.py with specified arguments. The original raw data comes from the Wu-Minn HCP 1200 subjects data release, and we use pre-processed data from ConnectomeDB. To learn more about the task information, please visit this link.

Please cite our paper if you find this code or our paper useful for your work:

@article{li2023interpretable,
  title={Interpretable Sparsification of Brain Graphs: Better Practices and Effective Designs for Graph Neural Networks},
  author={Li, Gaotang and Duda, Marlena and Zhang, Xiang and Koutra, Danai and Yan, Yujun},
  journal={arXiv preprint arXiv:2306.14375},
  year={2023}
}