A bidirectional Interpretable compound-protein interaction prediction framework based on cross attention

Scripts

1. Training model

(1) 5-fold cross validation

Usage

We provided two major scripts for training model:

-- ./preprocessing_data.py Converting SMILES strings to molecular graphs and constructing word dictionary for protein sequences.

-- ./train.py Training model.

Train model with the following command:

bash Run_Train.sh

The parameter 'scenario' in Run_Train.sh should be set to 'cross_validation'.

(2) cluster-cross validation

Usage

-- ./file_preparation.py Preparating files for clustering.

-- ./clustering.py Clustering for compounds and proteins.

-- ./split_dataset_according_clustering.py Spliting dataset into train/val/test set based on clustering.

-- ./preprocessing_data.py Converting SMILES strings to molecular graphs and constructing word dictionary for protein sequences.

-- ./train.py Training model.

Train model with the following command:

bash Run_Train.sh

The parameter 'scenario' in Run_Train.sh should be set to 'unknown_setting'.

2. Applying pre-trained model on test data

Usage

We provided two major scripts for predicting test data:

-- ./preprocessing_test_data.py Converting SMILES strings of the test data to molecular graphs and constructing word dictionary for protein sequences of the test data.

-- ./predict.py Make predictions on the test data.

Predict on the test data with the following command:

bash Run_Predicting.sh

Data

Four CPI datasets: BindingDB, DrugBank, GPCR and Davis

The datsets BindingDB.txt, DrugBank.txt, GPCR.txt and Davis.txt used in this study are provided at https://drive.google.com/drive/folders/1lfVZNHlpgdlBhozK_oeS1upU8NvpBOx6?usp=sharing

Cite：

Meng Wang, Jianmin Wang, Zhiwei Rong, Liuying Wang, Zhenyi Xu, Liuchao Zhang, Jia He, Shuang Li, Lei Cao, Yan Hou, Kang Li. A bidirectional interpretable compound-protein interaction prediction framework based on cross attention. Computers in Biology and Medicine, 2024, 172: 108239, https://doi.org/10.1016/j.compbiomed.2024.108239