Search

Community Structure-Activity Resource (CSAR) conducted a benchmark exercise to evaluate the current computational methods for protein design, ligand docking, and scoring/ranking. The exercise consisted of three phases. The first phase required the participants to identify and rank order which designed sequences were able to bind the small molecule digoxigenin. The second phase challenged the community to select a near-native pose of digoxigenin from a set of decoy poses for two of the designed proteins. The third phase investigated the ability of current methods to rank/score the binding affinity of 10 related steroids to one of the designed proteins (pKd = 4.1 to 6.7). We found that 11 of 13 groups were able to correctly select the sequence that bound digoxigenin, with most groups providing the correct three-dimensional structure for the backbone of the protein as well as all atoms of the active-site residues. Eleven of the 14 groups were able to select the appropriate pose from a set of plausible decoy poses. The ability to predict absolute binding affinities is still a difficult task, as 8 of 14 groups were able to correlate scores to affinity (Pearson-r > 0.7) of the designed protein for congeneric steroids and only 5 of 14 groups were able to correlate the ranks of the 10 related ligands (Spearman-ρ > 0.7).

The 2014 CSAR Benchmark Exercise was the last community-wide exercise that was conducted by the group at the University of Michigan, Ann Arbor. For this event, GlaxoSmithKline (GSK) donated unpublished crystal structures and affinity data from in-house projects. Three targets were used: tRNA (m1G37) methyltransferase (TrmD), Spleen Tyrosine Kinase (SYK), and Factor Xa (FXa). A particularly strong feature of the GSK data is its large size, which lends greater statistical significance to comparisons between different methods. In Phase 1 of the CSAR 2014 Exercise, participants were given several protein-ligand complexes and asked to identify the one near-native pose from among 200 decoys provided by CSAR. Though decoys were requested by the community, we found that they complicated our analysis. We could not discern whether poor predictions were failures of the chosen method or an incompatibility between the participant's method and the setup protocol we used. This problem is inherent to decoys, and we strongly advise against their use. In Phase 2, participants had to dock and rank/score a set of small molecules given only the SMILES strings of the ligands and a protein structure with a different ligand bound. Overall, docking was a success for most participants, much better in Phase 2 than in Phase 1. However, scoring was a greater challenge. No particular approach to docking and scoring had an edge, and successful methods included empirical, knowledge-based, machine-learning, shape-fitting, and even those with solvation and entropy terms. Several groups were successful in ranking TrmD and/or SYK, but ranking FXa ligands was intractable for all participants. Methods that were able to dock well across all submitted systems include MDock,1 Glide-XP,2 PLANTS,3 Wilma,4 Gold,5 SMINA,6 Glide-XP2/PELE,7 FlexX,8 and MedusaDock.9 In fact, the submission based on Glide-XP2/PELE7 cross-docked all ligands to many crystal structures, and it was particularly impressive to see success across an ensemble of protein structures for multiple targets. For scoring/ranking, submissions that showed statistically significant achievement include MDock1 using ITScore1,10 with a flexible-ligand term,11 SMINA6 using Autodock-Vina,12,13 FlexX8 using HYDE,14 and Glide-XP2 using XP DockScore2 with and without ROCS15 shape similarity.16 Of course, these results are for only three protein targets, and many more systems need to be investigated to truly identify which approaches are more successful than others. Furthermore, our exercise is not a competition.

The Community Structure-Activity Resource (CSAR) recently held its first blinded exercise based on data provided by Abbott, Vertex, and colleagues at the University of Michigan, Ann Arbor. A total of 20 research groups submitted results for the benchmark exercise where the goal was to compare different improvements for pose prediction, enrichment, and relative ranking of congeneric series of compounds. The exercise was built around blinded high-quality experimental data from four protein targets: LpxC, Urokinase, Chk1, and Erk2. Pose prediction proved to be the most straightforward task, and most methods were able to successfully reproduce binding poses when the crystal structure employed was co-crystallized with a ligand from the same chemical series. Multiple evaluation metrics were examined, and we found that RMSD and native contact metrics together provide a robust evaluation of the predicted poses. It was notable that most scoring functions underpredicted contacts between the hetero atoms (i.e., N, O, S, etc.) of the protein and ligand. Relative ranking was found to be the most difficult area for the methods, but many of the scoring functions were able to properly identify Urokinase actives from the inactives in the series. Lastly, we found that minimizing the protein and correcting histidine tautomeric states positively trended with low RMSD for pose prediction but minimizing the ligand negatively trended. Pregenerated ligand conformations performed better than those that were generated on the fly. Optimizing docking parameters and pretraining with the native ligand had a positive effect on the docking performance as did using restraints, substructure fitting, and shape fitting. Lastly, for both sampling and ranking scoring functions, the use of the empirical scoring function appeared to trend positively with the RMSD. Here, by combining the results of many methods, we hope to provide a statistically relevant evaluation and elucidate specific shortcomings of docking methodology for the community.

A major goal in drug design is the improvement of computational methods for docking and scoring. The Community Structure Activity Resource (CSAR) has collected several data sets from industry and added in-house data sets that may be used for this purpose ( www.csardock.org). CSAR has currently obtained data from Abbott, GlaxoSmithKline, and Vertex and is working on obtaining data from several others. Combined with our in-house projects, we are providing a data set consisting of 6 protein targets, 647 compounds with biological affinities, and 82 crystal structures. Multiple congeneric series are available for several targets with a few representative crystal structures of each of the series. These series generally contain a few inactive compounds, usually not available in the literature, to provide an upper bound to the affinity range. The affinity ranges are typically 3-4 orders of magnitude per series. For our in-house projects, we have had compounds synthesized for biological testing. Affinities were measured by Thermofluor, Octet RED, and isothermal titration calorimetry for the most soluble. This allows the direct comparison of the biological affinities for those compounds, providing a measure of the variance in the experimental affinity. It appears that there can be considerable variance in the absolute value of the affinity, making the prediction of the absolute value ill-defined. However, the relative rankings within the methods are much better, and this fits with the observation that predicting relative ranking is a more tractable problem computationally. For those in-house compounds, we also have measured the following physical properties: logD, logP, thermodynamic solubility, and pK(a). This data set also provides a substantial decoy set for each target consisting of diverse conformations covering the entire active site for all of the 58 CSAR-quality crystal structures. The CSAR data sets (CSAR-NRC HiQ and the 2012 release) provide substantial, publically available, curated data sets for use in parametrizing and validating docking and scoring methods.