GIFDTI

GIFDTI: Prediction of drug-target interactions based on global molecular and intermolecular interaction representation learning This repository contains the source code and the data.

GIFDTI

<div align="center"> <p><img src="model.jpg" width="600" /></p> </div>

Setup and dependencies

Dependencies:

• python 3.6
• pytorch >=1.2
• numpy
• sklearn
• tqdm
• tensorboardX
• prefetch_generator

Resources:

• README.md: this file.

• data: The datasets used in paper.

  • DrugBank2021.txt:
  • KIBA.txt:
  • Davis.txt
  • BindingDB In the directory of data, we now have the original data "DrugBank/KIBA/Davis.txt" as follows:

  Drug_ID Protein_ID Drug_SMILES Amino_acid_sequence interaction
  DB00303 P45059 [H][C@]12[C@@H]... MVKFNSSRKSGKSKKTIRKLT... 1
  DB00114 P19113 CC1=NC=C(COP(O)... MMEPEEYRERGREMVDYICQY... 1
  DB00117 P19113 N[C@@H](CC1=CNC... MMEPEEYRERGREMVDYICQY... 1
  ...
  ...
  ...
  DB00441 P48050 NC1=NC(=O)N(C=C... MHGHSRNGQAHVPRRKRRNRF... 0
  DB08532 O00341 FC1=CC=CC=C1C1=... MVPHAILARGRDVCRRNGLLI... 0
  
  

• dataset.py: data process.

• main.py: train and test the model under S1 setting.

• denovel.py: train and test the model under S2-s4 setting.

• hyperparameter.py: set the hyperparameter

• model.py: model architecture

Run:

python main.py