GPU supported AMBER TI module is in active development and is expected to be available in the AMBER16 release [30]. Conclusion Herein, we first repeated the calculation with the same data set as used in the original FEW [14] work which led to similar correlation R2 to the experiments. determined by the FEP Mapper tool. The uncorrected FEP-predicted relative binding affinities and cycle closure corrected energetics for the ligands in both neutral state and charged state are shown in Table S3. By fitting to the experimental results, a correlation coefficient R2 of 0.38 was found for the neutral ligand pair transformations and 0.49 for the charged ligand pairs. Interestingly, the cycle closure correction contributed no improvement to the correlation. Comparison of Schr?dinger FEP and AMBER TI workflows By taking the same input structures, the calculation results by using Schr?dinger FEP (corrected predictions) were highly correlated with the AMBER TI predictions (with linear extrapolation). They are almost equivalent with a correlation R2 = 0.80, RMSE = 0.64 kcal/mol at neutral state, and R2 = 0.96, RMSE = 0.30 kcal/mol when the charges and protonation states are corrected for all the ligands (the correlation of Amber TI and Schr?dinger FEP are shown in Fig. 2). Open in a separate window Fig. 2 Correlation of AMBER FEW TI prediction with Schr?dinger FEP for the relative binding affinities of ligand transformations at neutral state (Both plots showed the AMBER TI result using extrapolation and convergence method 1. Similar correlations were found by using no extrapolation or convergence method 2, shown in Fig. S2) The Schr?dinger FEP and AMBER TI workflows are then comparable except for the speed: for the AMBER TI workflow, it takes approximately 1 week to perform one transformation with TI calculation on a state-of-the-art computer cluster using 64 CPU cores (16 cores/node) per window, but it only takes a day or less to complete one Schr?dinger FEP calculation using 4 GPU cores per transformation. GPU supported AMBER TI module is in active development and is expected to be available in the AMBER16 release [30]. Conclusion Herein, we first repeated the calculation with the same data set as used in the original FEW [14] work which led to similar correlation R2 to the experiments. Then, we carefully examined the influences of using different force fields and control parameters, and further investigated the effect of protonation and tautomerization claims within the determined ligand binding affinity. Variance of the convergence methods in AMBER FEW makes negligible difference to the correlation of the prediction to experimental data. However, linear extrapolation slightly reduced the accuracy of the predictions. As expected, the AMBER ff12SB enhances the correlation R2 to the experiments from 0.29 to 0.41 (or from 0.35 to 0.42) compared to the ff99SB pressure field. Compared to the published predictions based on Element X inhibitors in their neutral state, the usage of right protonation claims boosted both AMBER TI and Schr?dinger FEP, where the result R2 correlation was improved up to 0.49 in Schr?dinger FEP and 0.73 in AMBER TI. Using the right tautomer state significantly reduced the prediction error, and corrected the rating between the example inhibitors transformation (e.g. L51a to L51b). We further benchmarked the AMBER TI in FEW with the Schr?dinger FEP+. To our knowledge, this is the 1st performance assessment of predictions between the AMBER FEW with the Schr?dinger FEP+. Even though AMBER TI calculation is definitely relatively sluggish, the accuracy of both methods is almost comparative. It proves the AMBER TI method can be useful for accurately determining relative binding affinity of chemically related, pharmaceutical-like compounds. Supplementary Material SIClick here to view.(816K, pdf) Acknowledgments We are grateful to Merck Study Laboratories (MRL) Postdoctoral Study Fellows System for monetary support provided by a fellowship (Y. H.). We say thanks to the AMBER FEW designers Nadine Homeyer and Holger Gohlke for useful help and discussions in building the workflows. We say thanks to the High Performance Computing (HPC) support at Merck. Abbreviations TIThermodynamic integrationFEPFree energy perturbationMM-GBSAMolecular mechanics-generalized given birth to surface areaMM-PBSAMolecular mechanics-Poisson Boltzmann surface areaLIELinear connection energyMCSSMaximum common substructure searchFEWFree-energy workflowsSBDDStructure-based drug designMDMolecular dynamics Footnotes Electronic supplementary material The online version of this article (doi:10.1007/s10822-016-9920-5) contains supplementary material, which is available to authorized users..To our knowledge, this is the first performance comparison of predictions between the AMBER FEW with the Schr?dinger FEP+. state are demonstrated in Table S3. By fitted to the experimental results, a correlation coefficient R2 of 0.38 was found for the neutral ligand pair transformations and 0.49 for the charged ligand pairs. Interestingly, the cycle closure correction contributed no improvement to the correlation. Assessment of Schr?dinger FEP and AMBER TI workflows By taking the same input structures, the calculation results by using Schr?dinger FEP (corrected predictions) were highly correlated with the AMBER TI predictions (with linear extrapolation). They may be almost equivalent having a correlation R2 = 0.80, RMSE = 0.64 kcal/mol at neutral state, and R2 = 0.96, RMSE = 0.30 kcal/mol when the charges and protonation states are corrected for all the ligands (the correlation of Amber TI and Schr?dinger FEP are shown in Fig. 2). Open in a separate windows Fig. 2 Correlation of AMBER FEW TI prediction with Schr?dinger FEP for the family member binding affinities of ligand transformations at neutral state (Both plots showed the AMBER TI result using extrapolation and convergence method 1. Related correlations were found by using no extrapolation or convergence method 2, demonstrated in Fig. S2) The Schr?dinger FEP and AMBER TI workflows are then comparable except for the rate: for the AMBER TI workflow, it takes approximately Clofilium tosylate 1 week to perform 1 transformation with TI calculation on a state-of-the-art computer cluster using 64 CPU cores (16 cores/node) per windows, but it only takes a day time or less to complete 1 Schr?dinger FEP calculation using 4 GPU cores per transformation. GPU supported AMBER TI module is in active development and AFX1 is expected to be available in the AMBER16 launch [30]. Summary Herein, we 1st repeated the calculation with the same data arranged as used in the original FEW [14] work which led to similar correlation R2 to the experiments. Then, we cautiously examined the influences of using different pressure fields and control guidelines, and further investigated the effect of protonation and tautomerization claims on the determined ligand binding affinity. Variance of the convergence methods in AMBER FEW makes negligible difference to the correlation of the prediction to experimental data. However, linear extrapolation slightly reduced the accuracy of the predictions. As expected, the AMBER ff12SB enhances the correlation R2 to the experiments from 0.29 to 0.41 (or from 0.35 to 0.42) compared to the ff99SB pressure field. Compared to the published predictions based on Element X inhibitors in their neutral state, the usage of correct protonation claims boosted both AMBER TI and Schr?dinger FEP, where the result R2 correlation was improved up to 0.49 in Schr?dinger FEP Clofilium tosylate and 0.73 in AMBER TI. Using the right tautomer state significantly reduced the prediction error, and corrected the rating between the example inhibitors transformation (e.g. L51a to L51b). We further benchmarked the AMBER TI in FEW with the Schr?dinger FEP+. To our knowledge, this is the 1st performance assessment of predictions between the AMBER FEW with the Schr?dinger FEP+. Even though AMBER TI calculation is relatively sluggish, the accuracy of both methods is almost comparative. It proves the AMBER TI method can be useful for accurately determining relative binding affinity of chemically related, pharmaceutical-like compounds. Supplementary Material SIClick here to see.(816K, pdf) Acknowledgments We are grateful to Merck Analysis Laboratories (MRL) Postdoctoral Analysis Fellows Plan for economic support supplied by a fellowship (Con. H.). We give thanks to the AMBER FEW programmers Nadine Homeyer and Holger Gohlke for beneficial help and conversations in building the workflows. We give thanks to the POWERFUL Processing (HPC) Clofilium tosylate support at Merck. Abbreviations TIThermodynamic integrationFEPFree energy perturbationMM-GBSAMolecular mechanics-generalized delivered surface area areaMM-PBSAMolecular mechanics-Poisson Boltzmann surface area areaLIELinear interaction.
Opioid, ??-