/TopoPharmDTI

Primary LanguagePython

TopoPharmDTI: Improving Interactions Prediction by Enhanced Deep Learning Representa-tion for Both Drug and Target Molecules

Model architecture

image

Setup and dependencies

dependencies:

  • python 3.9
  • pytorch >= 2.0.1
  • numpy
  • pandas
  • RDkit
  • pandas
  • pyg

Source codes:

  • cpi_model.py: TopoPharmDTI model file
  • smiles2graph.py: Generate drug features
  • smiles2subgraph.py: Generate drug subgraph features
  • protein_embedding.py: Generate protein features
  • train.py: train a TopoPharmDTI model.

Model and Datasets

Trained models and datasets is now available freely at https://drive.google.com/file/d/1SX8xc_jHDnGyPRKm6m6_oe_H9hXHkfyN/view?usp=sharing.

Run

  • generate drug and protein features：Generate two types of drug features through smiles2Graph.py and smiles2Subgraph.py, respectively
  • Generate protein features by running protein_embedding.py
  • begain train
    python train.py
  • inference:Load the trained model for testing on the testset
    python prediction.py

Case study

image