AutoDock-SS: Ligand-Based Virtual Screening Algorithm

Introduction

AutoDock-SS is a novel algorithm designed for ligand-based virtual screening. It leverages advanced computational techniques to efficiently screen large compound libraries against a reference ligand, facilitating the discovery of potential drug candidates. https://doi.org/10.1021/acs.jcim.4c00136

Features

• Efficient screening of large compound libraries
• Utilization of .pdbqt format for reference ligands
• Support for .sdf.gz compressed SDF file format for compound libraries
• Integration with AutoDock-GPU for enhanced performance

Installation

1. For AutoDock-GPU compilation, please refer to https://github.com/ccsb-scripps/AutoDock-GPU
2. Clone the AutoDock-SS repository:

   git clone https://github.com/NiBoyang/AutoDock-SS.git

3. Create a conda environment using the provided env.yaml file:

   conda env create -f env.yaml

4. Activate the environment:

   conda activate adss

Usage

1. Prepare your reference ligand in .pdbqt format and your compound library in a compressed .sdf.gz file.
2. Modify the common variables in adss_main.py to match your file paths and settings:

   autodock_gpu_exec_path = "" # directory of Autodock-GPU exec file
   lib_path = f'' # directory of your VS library
   lig_path = f'' # directory of your reference ligand
   path_of_scripts = "" # directory of Autodock-SS scripts

3. Run AutoDock-SS:

   python adss_main.py

Components

• adss_main.py: Main script to run the virtual screening process.
• env.yaml: Conda environment file with necessary dependencies.
• ligandprep.py: Script for ligand preparation. Called in the main file.
• roc_auc.py: Script for evaluating the performance of the screening. Called in the main file.
• Utility scripts: geometry_utils.py, grid_map_utils.py, file_utils.py for various computational tasks.
• DA.ipynb: A Jupyter notebook for data analysis.
• cal_score.py: Script for normalizing the scores if they were not correctly calculated by AutoDock-SS. You need to name your reference ligand(s) end with the suffix '_lig'. For example, if a reference ligand is named 'abc', then rename it as 'abc_lig' and put it in your whole library. After running the main file, this file could be used to re-calculate the normalized scores if needed.

Example Results

The Result_Examples directory contains example outputs from the AutoDock-SS algorithm, showcasing its capabilities.

Note

The grid_map_utils.py is the most essential file. You can generate ligand-based grid maps by refering to this file, and feed the map files to AutoDock.

License

AutoDock-SS is released under MIT License.

Acknowledgements

Special thanks to all contributors and the open-source community for their support and contributions to this project.