High-Quality Dataset of Protein-Bound Ligand Conformations and Its Application to Benchmarking Conformer Ensemble Generators

Abstract

We developed a cheminformatics pipeline for the fully automated selection and extraction of high-quality protein-bound ligand conformations from X-ray structural data. The pipeline evaluates the validity and accuracy of the 3D structures of small molecules according to multiple criteria, including their fit to the electron density and their physicochemical and structural properties. Using this approach, we compiled two high-quality datasets from the Protein Data Bank (PDB): a comprehensive dataset and a diversified subset of 4626 and 2912 structures, respectively. The datasets were applied to benchmarking seven freely available conformer ensemble generators: Balloon (two different algorithms), the RDKit standard conformer ensemble generator, the Experimental-Torsion basic Knowledge Distance Geometry (ETKDG) algorithm, Confab, Frog2 and Multiconf-DOCK. Substantial differences in the performance of the individual algorithms were observed, with RDKit and ETKDG generally achieving a favorable balance of accuracy, ensemble size and runtime. The Platinum datasets are available for download from http://www.zbh.uni-hamburg.de/platinum_dataset .

BCL::Conf – Improved Open-Source Knowledge-Based Conformation Sampling using the Crystallographic Open Database

“…This is a comparison to the results displayed inTable 3 in Friedrich et al20173 ,Table 2in Friedrich et al 2017 4 , and Table 2 in Cole et al 2018 5 . All results for non-BCL methods are as reported previously.…”

Fast, Efficient Fragment-Based Coordinate Generation for Open Babel

“…The relatively high RMSD distribution is largely due to differences in conformations between the stochastic single pose considered here and an ensemble of diverse conformers needed to find low-RMSD geometries. [35,36,37] In addition to the mean torsion angle error between the generated geometries and the experimental geometries, we computed the torsion fingerprint deviation (TFD) [34] using RDKit, as an established metric for comparison of torsional errors. The metric ignores hydrogen atoms and minimizes effects of dihedral angles with multiple symmetric atoms.…”

BCL::Conf – Improved Open-Source Knowledge-Based Conformation Sampling using the Crystallographic Open Database

“…2017 3 , Table 2 in Friedrich et al 2017 4 , and Our previously published conformer generator 9 , hereon referred to as BCL::Conf2016, with the use of rotamers derived from the CSD, more effectively recovers crystallographic ligand-binding conformations seen in the PDB than other freely available software, with the possible exception of Confab (though Confab fails to produce conformers for a large fraction of the dataset, and thus results are difficult to interpret; see Table 2 note (b)). With the improvements made to BCL::Conf discussed above, the BCL more effectively recovers crystallographic ligand-binding conformations than all competing free and commercial methods (Table 2).…”

“…Average time to generate BCL conformers with increasing numbers of iterations and requested conformers. RMSD measured with the RDKit Symmetry RMSD metric.Two of the academic software packages demonstrate faster performance than BCL::Conf -Confab and Frog2; however, these methods fail to generate conformations for a large percentage of compounds in the dataset3 .Conclusions, limitations, and future directionsHere we have presented substantial improvements to the BCL::Conf algorithm. The BCL is an open-source, free for academic use cheminformatics toolkit integrating classic tools (e.g.…”

“…For comparison of conformers with the native conformers, we used Symmetry RMSD, as implemented in BCL. BCL's implementation of Symmetry RMSD considers isometric atoms equivalent, and allows for swapping of terminal single/double bonds to identical elements, and is thus comparable with prior studies.As stated in the Friedrich et al 20173 paper, "The term 'accuracy of a conformational ensemble' usually refers to the RMSD [Å], calculated for the best-fitting conformer of an ensemble compared to the experimentally observed conformer [...] the RMSD remains the de facto standard in benchmarking conformer ensemble generators." Therefore, the benchmark below is a standardized evaluation against leading conformer generator software packages.…”