1. I used three datasets, one of which was the original MSI(dataset/MSI), One is the MSI of a screened drug-disease pair(dataset/MSI_new),The other is the data I collated myself(dataset/My)
2. to MSI_new,The similarity calculated by the code (similarity_graph.txt，DREAMwalk/code/gengerate_similarity_net.py)， 
    and the similarity you provided (similarity_graph_2)
3. MSI_new and My data sets are divided by randomly selecting negative samples with the same number of positive samples(dataset/MSI_new/split_data.py）
   to MSI using what you mentioned in your pape"Negative drug-disease pairs of equal number of positive pairs were randomly sampled from the
   network. ten-times ten-fold cross-validation (CV) setting was adopted, and negative sampling was conducted in a way that there are no
    overlapping samples between the ten CV sets."（dataset/MSI/data_fold/code/run.py）


lq_code
   dataset( used dataset)    
       Drugbank
       Mesh
       MSI
       MSI_new
       My
   DREAMwalk 
       code(main code)
       data (Data generated by the intermediate process)     
       model(Generated model)
       result(result)