About
High-throughput identification of potential ligand-binding pockets is essential for structure-based drug discovery and for anticipating off-target interactions. In this study, we developed an automated workflow to analyze 220,019 protein structures from the Protein Data Bank (PDB) using Fpocket, obtaining 219,762 valid outputs. Fpocket detected 12,360,014 pockets in total, from which 93,971 pockets with PocketScore and DrugScore ≥ 0.5 were retained as having predicted ligandability. All corresponding protein structures were converted into PDBQT format for immediate use with AutoDock Vina, and comprehensive pocket metadata were extracted, including pocket centers, box dimensions, and the identities and 3D coordinates of all amino acid residues lining each pocket. These annotations allow the database to be used not only with Vina but also with other docking engines and pocket-based analyses. The resulting PDBPockets resource provides a robust foundation for large-scale inverse docking, structure-based virtual screening, therapeutic repurposing opportunities, and global exploration of protein–ligand interaction space.
Download
The program to generate the virtual library of fragments (FragVLib) is Open Source software that is available at no charge. It is supplied as is, without warranty, and may be freely used and distributed under the terms of the Gnu General Public License (GPL).
To download the PDBPockets database, follow this link: PDBPockets.tar.gz
