Vencent-Won / SGGRL

The implementation of the paper "Multi-Modal Representation Learning for Molecular Property Prediction: Sequence, Graph, Geometry".

SGGRL: Multi-Modal Representation Learning for Molecular Property Prediction: Sequence, Graph, Geometry

Apology

Dear Colleagues,

I apologize for the delay in responding to your queries. I've been preoccupied with my regular workload. I'll address these issues as soon as possible. Sorry for any inconvenience this may have caused.

Best regards,

Zeyu Wang

Framework

Enviroment

• paddle-bfloat==0.1.7
• paddlepaddle==2.5.1
• torch==1.13.0
• torch-cluster==1.6.0+pt113cu117
• torch-geometric==2.2.0
• torch-scatter==2.1.0+pt113cu117
• torch-sparse==0.6.15+pt113cu117
• torch-spline-conv==1.2.1+pt113cu117
• rdkit==2023.3.1

Usage:

1. Process Data

python build_corpus.py --in_path {data_path} --out_path {save_path}
python build_vocab.py --corpus_path {corpus_path} --out_path {save_path}
python data_3d.py --dataset {dataset name}

2. Molecular Property Prediction

python main.py --dataset {dataset name} --task_type {reg/class}

Citation

Wang Z, Jiang T, Wang J, et al. Multi-Modal Representation Learning for Molecular Property Prediction: Sequence, Graph, Geometry[J]. arXiv preprint arXiv:2401.03369, 2024.