AMDGT: attention aware multi-modal learning using dual graph transformer for drug-disease associations prediction

Requirements:

• python 3.9.13
• cudatoolkit 11.3.1
• pytorch 1.10.0
• dgl 0.9.0
• networkx 2.8.4
• numpy 1.23.1
• scikit-learn 0.24.2

Data:

The data files needed to run the model, which contain C-dataset and F-dataset.

• Drug_mol2vec: The mol2vec embeddings for drugs to construct the association network
• DrugFingerprint, DrugGIP: The similarity measurements of drugs to construct the similarity network
• DiseaseFeature: The disease embeddings to construct the association network
• DiseasePS, DiseaseGIP: The similarity measurements of diseases to construct the similarity network
• Protein_ESM: The ESM-2 embeddings for proteins to construct the association network
• DrugDiseaseAssociationNumber: The known drug disease associations
• DrugProteinAssociationNumber: The known drug protein associations
• ProteinDiseaseAssociationNumber: The known disease protein associations

Code:

• data_preprocess.py: Methods of data processing
• metric.py: Metrics calculation
• train_DDA.py: Train the model

Usage:

Execute python train_DDA.py