Leveraging binding pose metadynamics to optimise target fishing predictions for three diverse ligands and their true targets.

Computational target fishing plays an important role in target identification, particularly in drug discovery campaigns utilizing phenotypic screening. Numerous approaches exist to predict potential targets for a given ligand, but true targets may be inconsistently ranked. More advanced simulation methods may provide benefit in such cases by reranking these initial predictions. We evaluated the ability of binding pose metadynamics to improve the predicted rankings for three diverse ligands and their six true targets. Initial predictions using pharmacophore mapping showed no true targets ranked in the top 50 and two targets each ranked within the 50-100, 100-150, and 250-300 ranges respectively. Following binding pose metadynamics, ranking of true targets improved for four out of the six targets and included the highest ranked predictions overall, while rankings deteriorated for two targets. The revised rankings predicted two true targets ranked within the top 50, and one target each within the 50-100, 100-150, 150-200, and 200-250 ranges respectively. The findings of this study demonstrate that binding pose metadynamics may be of benefit in refining initial predictions from structure-based target fishing algorithms, thereby improving the efficiency of the target identification process in drug discovery efforts.

Consensus scoring evaluated using the GPCR-Bench dataset: Reconsidering the role of MM/GBSA.

The recent availability of large numbers of GPCR crystal structures has provided an unprecedented opportunity to evaluate their performance in virtual screening protocols using established benchmarking datasets. In this study, we evaluated the ability of MM/GBSA in consensus scoring-based virtual screening enrichment together with nine classical scoring functions, using the GPCR-Bench dataset consisting of 24 GPCR crystal structures and 254,646 actives and decoys. While the performance of consensus scoring was modest overall, combinations which included MM/GBSA performed relatively well compared to combinations of classical scoring functions. Combinations of MM/GBSA and good-performing scoring functions provided the highest proportion of improvements, with improvements observed in 32% and 19% of all combinations across all targets at the EF1% and EF5% levels respectively. Combinations of MM/GBSA and poor-performing scoring functions still outperformed classical scoring functions, with improvements observed in 26% and 17% of all combinations at the EF1% and EF5% levels. In comparison, only 14-22% and 6-11% of combinations of classical scoring functions produced improvements at EF1% and EF5% respectively. Efforts to improve performance by increasing the number of scoring functions in consensus scoring to three were mostly ineffective. We also observed that consensus scoring performed better for individual scoring functions possessing initially low enrichment factors, potentially implying their benefits are more relevant in such scenarios. Overall, this study demonstrated the first implementation of MM/GBSA in consensus scoring using the GPCR-Bench dataset and could provide a valuable benchmark of the performance of MM/GBSA in comparison to classical scoring functions in consensus scoring for GPCRs.

Benchmarking the performance of MM/PBSA in virtual screening enrichment using the GPCR-Bench dataset.

Recent breakthroughs in G protein-coupled receptor (GPCR) crystallography and the subsequent increase in number of solved GPCR structures has allowed for the unprecedented opportunity to utilize their experimental structures for structure-based drug discovery applications. As virtual screening represents one of the primary computational methods used for the discovery of novel leads, the GPCR-Bench dataset was created to facilitate comparison among various virtual screening protocols. In this study, we have benchmarked the performance of Molecular Mechanics/Poisson-Boltzmann Surface Area (MM/PBSA) in improving virtual screening enrichment in comparison to docking with Glide, using the entire GPCR-Bench dataset of 24 GPCR targets and 254,646 actives and decoys. Reranking the top 10% of the docked dataset using MM/PBSA resulted in improvements for six targets at EF1% and nine targets at EF5%, with the gains in enrichment being more pronounced at the EF1% level. We additionally assessed the utility of rescoring the top ten poses from docking and the ability of short MD simulations to refine the binding poses prior to MM/PBSA calculations. There was no clear trend of the benefit observed in both cases, suggesting that utilizing a single energy minimized structure for MM/PBSA calculations may be the most computationally efficient approach in virtual screening. Overall, the performance of MM/PBSA rescoring in improving virtual screening enrichment obtained from docking of the GPCR-Bench dataset was found to be relatively modest and target-specific, highlighting the need for validation of MM/PBSA-based protocols prior to prospective use.

Evaluating the performance of MM/PBSA for binding affinity prediction using class A GPCR crystal structures.

The recent expansion of GPCR crystal structures provides the opportunity to assess the performance of structure-based drug design methods for the GPCR superfamily. Molecular Mechanics/Poisson-Boltzmann Surface Area (MM/PBSA)-based methods are commonly used for binding affinity prediction, as they provide an intermediate compromise of speed and accuracy between the empirical scoring functions used in docking and more robust free energy perturbation methods. In this study, we systematically assessed the performance of MM/PBSA in predicting experimental binding free energies using twenty Class A GPCR crystal structures and 934 known ligands. Correlations between predicted and experimental binding free energies varied significantly between individual targets, ranging from r = - 0.334 in the inactive-state CB1 cannabinoid receptor to r = 0.781 in the active-state CB1 cannabinoid receptor, while average correlation across all twenty targets was relatively poor (r = 0.183). MM/PBSA provided better predictions of binding free energies compared to docking scores in eight out of the twenty GPCR targets while performing worse for four targets. MM/PBSA binding affinity predictions calculated using a single, energy minimized structure provided comparable predictions to sampling from molecular dynamics simulations and may be more efficient when computational cost becomes restrictive. Additionally, we observed that restricting MM/PBSA calculations to ligands with a high degree of structural similarity to the crystal structure ligands improved performance in several cases. In conclusion, while MM/PBSA remains a valuable tool for GPCR structure-based drug design, its performance in predicting the binding free energies of GPCR ligands remains highly system-specific as demonstrated in a subset of twenty Class A GPCRs, and validation of MM/PBSA-based methods for each individual case is recommended before prospective use.