A workflow to create a high-quality protein–ligand binding dataset for training, validation, and prediction tasks

2025/01/01 by Yingze Wang, Kunyang Sun, Jie Li +6 · 1 voice · 1 citation
Biochemistry, Genetics and Molecular Biology · Computer Science · #Bioinformatics and Genomic Networks #Computational Drug Discovery Methods #Protein Structure and Dynamics

paper · pdf · doi:10.1039/d4dd00357h

openalex publication_date 2025/01/01 · openalex created_date 2025/03/19 · openalex updated_date 2026/08/01

Abstract

Development of scoring functions (SFs) used to predict protein-ligand binding energies requires high-quality 3D structures and binding assay data for training and testing their parameters. In this work, we show that one of the widely-used datasets, PDBbind, suffers from several common structural artifacts of both proteins and ligands, which may compromise the accuracy, reliability, and generalizability of the resulting SFs. Therefore, we have developed a series of algorithms organized in a semi-automated workflow, HiQBind-WF, that curates non-covalent protein-ligand datasets to fix these problems. We also used this workflow to create an independent data set, HiQBind, by matching binding free energies from various sources including BioLiP, Binding MOAD and BindingDB with co-crystalized ligand-protein complexes from the PDB. The resulting HiQBind workflow and dataset are designed to ensure reproducibility and to minimize human intervention, while also being open-source to foster transparency in the improvements made to this important resource for the biology and drug discovery communities.

Citations

• A Smoothed Backbone-Dependent Rotamer Library for Proteins Derived from Adaptive Kernel Density Estimates and Regressions
• Development and validation of a genetic algorithm for flexible docking 1 1Edited by F. E. Cohen
• AutoDock Vina: Improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading
• OpenMM 8: Molecular Dynamics Simulation with Machine Learning Potentials
• AutoDock4 and AutoDockTools4: Automated docking with selective receptor flexibility
• Leak Proof PDBBind: A Reorganized Dataset of Protein-Ligand Complexes for More Generalizable Binding Affinity Prediction

Cited by

• Can AI-predicted complexes teach machine learning to compute drug binding affinity?

Discussions

• In our next featured article, Head-Gordon et al. introduce the HiQBind-WF workflow which addresses common artefacts in PDBbind data. Read the open access paper here: doi.org/10.1039/D4DD... Find HiQBi [bsky, 1 points, 0 comments]

Related

• A Workflow to Create a High-Quality Protein-Ligand Binding Dataset for Training, Validation, and Prediction Tasks
• Leak Proof PDBBind: A Reorganized Dataset of Protein-Ligand Complexes for More Generalizable Binding Affinity Prediction
• DualBind: A Dual-Loss Framework for Protein-Ligand Binding Affinity Prediction
• Comparative evaluation of methods for the prediction of protein–ligand binding sites
• Multi-task Bioassay Pre-training for Protein-ligand Binding Affinity Prediction