Datasets for Evaluation of Small-Molecule Binding Site Prediction Methods on Membrane-Embedded Protein Interfaces

The dataset contains PDB files of protein-ligand complexes for two major membrane protein classes - GPCRs and ion channels. In addition, it contains PDB files for a subset of 420 protein-ligand structures from the PDBBind dataset.

For each GPCR or ion channel a directory with the following content is provided: a PDB file downloaded from the OPM database (https://opm.phar.umich.edu) and a 'cleaned' PDB file in which all ligands, ions or lipids, except for the small molecule whose binding site should be predicted, were removed.

For each protein in the BDBBind set a directory with the following content is provided: a PDB file of the protein, a PDB and MOL2 file of the ligand and a PDB file of the combined protein-ligand complex.

A Scripts directory contains two Python scripts which were used to download PDB files from OPM database and clean them from all unwanted ligand molecules.