PDBScan22: Protein-Ligand Binding Sites and Redocking Results

322,051 macromolecule–ligand binding sites from the PDBScan22 dataset, with a high-quality subset of 21,355 sites, curated by researchers at Universität Hamburg. The repository includes calculated ligand and binding site properties, structure quality descriptors, and results from redocking experiments using JAMDA and AutoDock Vina. All data is provided as CSV files compliant with the CSV on the web standard.

Use Cases

• Benchmarking docking algorithms based on RMSD scores and success rates provided in the results files.
• Analyzing relationships between ligand properties (e.g., molecular weight) and docking performance.
• Filtering protein-ligand complexes for model training based on structure quality descriptors like crystal resolution.
• Studying binding site characteristics using calculated properties such as volume.

Strengths

• The PDBScan22 dataset contains 322,051 binding sites, providing a large-scale resource.
• A high-quality subset of 21,355 sites passes explicit structure quality filters.
• Data includes docking results from two distinct workflows (JAMDA and AutoDock Vina) for comparison.
• Files are structured with metadata JSON descriptions compliant with the CSV on the web standard.

Limitations

• Row count for individual files is unknown, which may limit suitability assessment.
• Some CSV files may contain missing values in columns other than the primary key identifiers.
• Last update date is unknown; freshness unverified.

Provenance

Source

Supplementary data for the journal article 'Redocking the PDB' by Flachsenberg et al.

Collection Method

Derived from the Protein Data Bank (PDB), processed with structure preparation and docking workflows.