wzy-Sarah / ZeroDDI

ZeroDDI: A Zero-Shot Drug-Drug Interaction Event Prediction Method with Semantic Enhanced Learning and Dual-Modal Uniform Alignmen

ZeroDDI

ZeroDDI: A Zero-Shot Drug-Drug Interaction Event Prediction Method with Semantic Enhanced Learning and Dual-Modal Uniform Alignment

Accepted by IJCAI2024

The authors are Ziyan Wang, Zhankun Xiong, Feng Huang, Xuan Liu, Wen Zhang.

1 install

create a conda virtual env: conda create -n name python=3.8

The required libraries are:

numpy

torch 1.11.0+cu113

addict

yapf

sklearn

pandas

torch_geometric 2.1.0

rdkit

tensorflow

deepchem

networkx

transformers>=4.26

matplotlib

sacremoses

bs4

lxml

2 Usage

2.1 dataset

Due to space limitations, we compressed the dataset. You can unzip all xxx.zip data in its fold.

2.2 Training ZeroDDI

There are three folds in /data/DrugBank5.1.9/

For example, the data of fold2 is in zsl2/ and gzsl2/

python main.py --config configs/zeroddi.py or python main.py --config configs/zeroddi_fold2.py

You can also create our own config python file for different datasets or models.

2.3 Testing ZeroDDI

After training, the parameters of models are saved in ./work_dirs/

Then, you can test the model by:

python main.py --config configs/zeroddi.py --zsl_para work_dirs/zeroddi/model_parameter/zsl_model_best_epoch100_seed42.pkl --gzsl_para work_dirs/zeroddi/model_parameter/gzsl_model_best_epoch100_seed42.pkl