HpyerAttentionDTI

HyperAttentionDTI: improving drug–protein interaction prediction by sequence-based deep learning with attention mechanism This repository contains the source code and the data.

HpyerAttentionDTI

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.

  • DrugBank.txt:
  • KIBA.txt:
  • Davis.txt 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.
• HpyerAttentionDTI_main.py: train and test the model.
• hyperparameter.py: set the hyperparameter of HpyerAttentionDTI
• model.py: HpyerAttentionDTA model architecture
• pytorchtools: early stopping

Run:

python HpyerAttentionDTI_main.py