Dynamic category-sensitive hypergraph inferring and homo-heterogeneous neighbor feature learning for drug-related microbe prediction
pytroch == 2.1.0
matplotlib == 3.8.0
numpy == 1.26.0
python == 3.9.18
data_process.py : Processing drug and microbial similarities and associations, forming embeddings, adjacency matrices, etc.
early_stoppoing.py : In order to save better parameters for the model
parameters.py : Hyperparameters of the model
tools4roc_pr.py : Evaluate the model
train.py : train the model
Model517.py ,Transformer.py ,model_all.py : Define the model
ST1.slsx : Top 20 candidates for every drug
drug_drug_interaction_adj.txt:Interactions between drugs
drugsimilarity.zip：Similarities between drugs
microbe_microbe_similarity:Similarities between microbes
net1.mat:Adjacency matrix of drug and microbe heterogeneous graph
1.data_process.py
2.train.py
Before running train.py, you need to create two folders, best_parameter and result, to store the model's parameters and training results