DiffBind

Python implementation of a denoising diffusion probabilistic model (DDPM) for predicting protein-ligand binding affinity using equivariant neural networks.

Overview

DiffBind leverages cutting-edge diffusion models and equivariant neural networks to predict the binding affinity between proteins and small molecules. This approach combines:

• Denoising Diffusion Probabilistic Models (DDPM) for generating protein-ligand binding poses
• Equivariant Neural Networks to respect 3D rotational and translational symmetries
• BindingMOAD Dataset for training and evaluation

Key Features

• 🧬 Equivariant molecular representations
• ⚗️ Diffusion-based binding pose generation
• 🎯 Binding affinity prediction
• 📊 Comprehensive evaluation metrics
• 🔬 Integration with RDKit and Biopython

Setting Up the Environment

To run the code in this repository, you'll need to create a Conda environment with specific dependencies. Below are the steps to set up the environment:

Creating a Conda Environment

1. Create a New Conda Environment:
   This command creates a new Conda environment named diff_bind.

   conda create -n diff_bind

2. Activate the Environment:
   Once the environment is created, activate it using:

   conda activate diff_bind

Installing Dependencies

Install the necessary dependencies by running the following commands in your activated environment:

• PyTorch with CUDA Toolkit:

  conda install pytorch cudatoolkit=10.2 -c pytorch

• PyTorch Lightning:

  conda install -c conda-forge pytorch-lightning

• Weights & Biases (wandb):

  conda install -c conda-forge wandb

• RDKit:

  conda install -c conda-forge rdkit

• Biopython:

  conda install -c conda-forge biopython

• ImageIO:

  conda install -c conda-forge imageio

• SciPy:

  conda install -c anaconda scipy

• PyTorch-Scatter:

  conda install -c pyg pytorch-scatter

• OpenBabel:

  conda install -c conda-forge openbabel

Final Steps

After installing all dependencies, your environment is ready to run the code in this repository.

Binding MOAD

This section describes how to prepare the Binding MOAD dataset for processing.

Data preparation

1. Download the Dataset: Run the following commands in your terminal to download the necessary files:

   wget http://www.bindingmoad.org/files/biou/every_part_a.zip
   wget http://www.bindingmoad.org/files/biou/every_part_b.zip
   wget http://www.bindingmoad.org/files/csv/every.csv
   
   unzip every_part_a.zip
   unzip every_part_b.zip

2. Process the raw data: Use the provided Python script to process the raw data. Replace '<bindingmoad_dir>' with the path to the directory where you've downloaded the Binding MOAD files:

   python -W ignore process_bindingmoad.py <bindingmoad_dir>

   • Optional: To create a dataset with only C-alpha (Cα) pocket representation, add the '--ca_only' flag to the command:

     python -W ignore process_bindingmoad.py <bindingmoad_dir> --ca_only

     This flag will configure the dataset to focus on the C-alpha atoms in the protein structure.