Pharmacophore searching: A potential solution for correcting unknown ligands (UNK) labelling errors in Protein Data Bank (PDB’S)

Abstract

The Protein Data Bank (PDB) is the single most important repository of structural data for proteins and other biologically relevant molecules. Therefore, it is critically important to keep the PDB data, error-free as much as possible. In this study, we have critically examined PDB structures of 292 protein molecules which have been deposited in the repository along with potentially incorrect ligands labelled as Unknown ligands (UNK). Pharmacophores were generated for all the protein structures by using Discovery Studio Visualizer (DSV) and Accelrys, Catalyst®. The generated pharmacophores were subjected to the database search containing the reported ligand. Ligands obtained through Pharmacophore searching were then checked for fitting the observed electron density map by using Coot®. The predicted ligands obtained via Pharmacophore searching fitted well with the observed electron density map, in comparison to the ligands reported in the PDB's. Based on our study we have learned that till may 2016, among 292 submitted structures in the PDB, at least 20 structures have ligands with a clear electron density but have been incorrectly labelled as unknown ligands (UNK). We have demonstrated that Pharmacophore searching and Coot® can provide potential help to find suitable known ligands for these protein structures, the former for ligand search and the latter for electron density analysis. The use of these two techniques can facilitate the quick and reliable labelling of ligands where the electron density map serves as a reference.