Introduction

This repository contains the implementation of the DeepDOX1 model for predicting protein-ligand binding affinities. The model leverages deep learning techniques to analyze molecular structures and predict their interactions.

Requirements

• Python 3.8+
• cuda 12.4
• PyTorch 2.4.1+cu124
• RDKit
• NumPy
• Pandas

The easiest way to install the required packages is to create environment with GPU-enabled version:

conda env create -f environment.yml
conda activate DeepDOX

Training & Testing

To train the DeepDOX1 model, run the following command:

cd ./src
python Train.py

To test the DeepDOX1 model, run the following command:

cd ./src
python Test_all.py

To Use the DeepDOX1 model, run the following command:

cd ./src
python Predict.py [model].pt [Datadir] [True/False]

Citation

Contact

For any questions or issues, please contact :