Skip to content

Latest commit

 

History

History
73 lines (60 loc) · 3.63 KB

README.md

File metadata and controls

73 lines (60 loc) · 3.63 KB
Raw

Decoupled peak property learning for efficient and interpretable ECD spectra prediction

The official code for "Decoupled peak property learning for efficient and interpretable ECD spectra prediction" submitted to Nature Computational Science. Here we publish the inference code of ECDFormer. The training code & ECD spectra dataset will be released after our paper is accepted. If you like our project, please give us a star ⭐ on GitHub for latest update.

arXiv License Data License

Data Preparation

For training and inference, please download and put the descriptor_all_column.npy into the folder utils/

utils/descriptor_all_column.npy

We will release the CMCDS dataset for training procedure once our paper is accepted.

🛠️ Requirements and Installation

  • Python == 3.8
  • Pytorch == 1.13.1
  • CUDA Version == 11.7
  • torch_geometric, troch-scatter, torch-sparse, torch-cluster, torch-spline-conv
  • Install required packages:
git clone [email protected]:HowardLi1984/ECDFormer.git
cd ECDFormer
pip install -r requirements.txt

PS: you can follow this link for faster torch_geometric install

## First install these related packages
pip install torch-scatter -f https://pytorch-geometric.com/whl/torch-${TORCH}+${CUDA}.html
pip install torch-sparse -f https://pytorch-geometric.com/whl/torch-${TORCH}+${CUDA}.html
pip install torch-cluster -f https://pytorch-geometric.com/whl/torch-${TORCH}+${CUDA}.html
pip install torch-spline-conv -f https://pytorch-geometric.com/whl/torch-${TORCH}+${CUDA}.html

## Then install the torch-geometric package
pip install torch-geometric

🗝️ Inferencing

The inferencing instruction is in main_func_pos.py.

CUDA_VISIBLE_DEVICES=0 python main_func_pos.py --model_name gnn_allthree --batch_size 256 --emb_dim 128 --epochs 1000 --lr 1e-3 --mode Real --visual_epoch 400

🚀 Main Results

Quantitively, we propose the experimental results on our ECDFormer framework and the corresponding baselines including machine learning models and deep learning models. Focusing on peak property prediction, our ECDFormer model surpasses baselines under all evaluation metrics.

ECD spectra predictions on natural products with pharmaceutical effects from recent journals demonstrate the effectiveness and generalization ability of our ECDFormer.

✏️ Citation

If you find our paper and code useful in your research, please consider giving a star ⭐ and citation 📝.

@article{li2024decoupled,
  title={Decoupled peak property learning for efficient and interpretable ECD spectra prediction},
  author={Li, Hao and Long, Da and Yuan, Li and Wang, Yu and Tian, Yonghong and Wang, Xinchang and Mo, Fanyang},
  year={2024}
}