Benchmark_VS

This repository contains the code and notebook to reproduce the analyses presented in the paper "Integrating Machine Learning-Based Pose Sampling with Established Scoring Functions for Virtual Screening"

Structure

Code

The utils/analysis_utils.py contains the code to:

• Calculate the virtual screening (VS) performance
• Extract best poses of each VS setup from docking pose files
• Generate PoseBusters check reports for the docking poses
• Calculate protein-ligand interaction fingerprint (PLIF) similarity between docked and reference molecules
• Calculate Morgan 2-based Tanimoto similarity between docked and reference molecules

Data

The input data used for the analysis experiments include:

• dudez: input molecules as SMILES strings provided in DUDE-Z for each target.
• docking_poses: docking poses with DiffDock-L and AutoDock Vina for 43 DUDE-Z targets.
• plif: the reference ligands collected with SIENA and the protein-ligand interaction fingerprints generated with ProLIF for docked compounds and reference ligands for each target.
• posebusters: the pose validity check results generated with PoseBusters for all docking poses for each target.

The above data can be downloaded from zenodo. The data can be then placed in data such that you have the paths: data/docking_poses, data/dudez, data/plif, and data/posebusters. This setup enables the execution of the analysis notebook in notebooks/ folder.

Setup Environment

Install python 3.10 in the virtual environment of choice and install the required packages noted in requirements.txt.

Running the analysis experiments

Run the notebook notebooks/benchmark_vs.ipynb to reproduce all the results presented in the paper.

The output files and figures will be placed in data/analysis by default. This is customizable.

Citing us

@article{vu2025integrating,
  title={Integrating Machine Learning-Based Pose Sampling with Established Scoring Functions for Virtual Screening},
  author={Vu, Thi Ngoc Lan and Fooladi, Hosein and Kirchmair, Johannes},
  journal={Journal of Chemical Information and Modeling},
  volume={65},
  number={10},
  pages={4833–4843},
  year={2025},
  publisher={ACS Publications}
}