Cargando…

Computational Calorimetry: High-Precision Calculation of Host–Guest Binding Thermodynamics

[Image: see text] We present a strategy for carrying out high-precision calculations of binding free energy and binding enthalpy values from molecular dynamics simulations with explicit solvent. The approach is used to calculate the thermodynamic profiles for binding of nine small molecule guests to...

Descripción completa

Detalles Bibliográficos
Autores principales: Henriksen, Niel M., Fenley, Andrew T., Gilson, Michael K.
Formato: Online Artículo Texto
Lenguaje:English
Publicado: American Chemical Society 2015
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4614838/
https://www.ncbi.nlm.nih.gov/pubmed/26523125
http://dx.doi.org/10.1021/acs.jctc.5b00405
Descripción
Sumario:[Image: see text] We present a strategy for carrying out high-precision calculations of binding free energy and binding enthalpy values from molecular dynamics simulations with explicit solvent. The approach is used to calculate the thermodynamic profiles for binding of nine small molecule guests to either the cucurbit[7]uril (CB7) or β-cyclodextrin (βCD) host. For these systems, calculations using commodity hardware can yield binding free energy and binding enthalpy values with a precision of ∼0.5 kcal/mol (95% CI) in a matter of days. Crucially, the self-consistency of the approach is established by calculating the binding enthalpy directly, via end point potential energy calculations, and indirectly, via the temperature dependence of the binding free energy, i.e., by the van’t Hoff equation. Excellent agreement between the direct and van’t Hoff methods is demonstrated for both host–guest systems and an ion-pair model system for which particularly well-converged results are attainable. Additionally, we find that hydrogen mass repartitioning allows marked acceleration of the calculations with no discernible cost in precision or accuracy. Finally, we provide guidance for accurately assessing numerical uncertainty of the results in settings where complex correlations in the time series can pose challenges to statistical analysis. The routine nature and high precision of these binding calculations opens the possibility of including measured binding thermodynamics as target data in force field optimization so that simulations may be used to reliably interpret experimental data and guide molecular design.