An industrial evaluation of proteochemometric modelling: predicting drug-target affinities for kinases

GraphDTA (https://github.com/thinng/GraphDTA) was adapted for the purpose of an industrial evaluation of deep learning proteochemometric models as part of the paper "An industrial evaluation of proteochemometric modelling: predicting drug-target affinities for kinases"

Installation

For python version 3.8, cuda version 11.3 and pytorch version 1.12.0, create a conda environment:

conda env create -f environment.yml

Activate the environment:

conda activate GraphDTAadapted

OR: Install Python libraries needed Install pytorch_geometric following instruction at https://github.com/rusty1s/pytorch_geometric Install rdkit: conda install -y -c conda-forge rdkit Install networkx: pip install networkx Install prettytable: pip install prettytable

Example usage

Training the model:

python training.py -h 
usage: training.py [-h] run data params

positional arguments:
  run         Name of the run.
  data        Name of the data.
  params      Config file.

python training.py  'test' 'BindingDB/RandomLigandSplit/Fold0' 'params.json'

Testing:

python testing.py -h
usage: testing.py [-h] model_dir run data params

positional arguments:
  model_dir   model directory
  run         Name of the run.
  data        Name of the data.
  params      Config file.