scholarly journals Bridging the experiment-calculation divide: machine learning corrections to redox potential calculations in implicit and explicit solvent models

2021 ◽  
Author(s):  
Eugen Hruska ◽  
Ariel Gale ◽  
Fang Liu
Keyword(s):  
Machine Learning ◽  
Redox Potential ◽  
Density Functional ◽  
Implicit Solvent ◽  
Systematic Bias ◽  
Redox Potentials ◽  
Solvent Models ◽  
Explicit Solvent ◽  
Solvent Model ◽  
Implicit And Explicit

Prediction of redox potentials is essential for catalysis and energy storage. Although density functional theory (DFT) calculations have enabled rapid redox potential predictions for numerous compounds, prominent errors persist compared to experimental measurements. In this work, we develop machine learning (ML) models to reduce the errors of redox potential calculations in both implicit and explicit solvent models. Training and testing of the ML correction models are based on the diverse ROP313 dataset with experimental redox potentials measured for organic and organometallic compounds in a variety of solvents. For the implicit solvent approach, our ML models can reduce both the systematic bias and the number of outliers. ML corrected redox potentials also demonstrate less sensitivity to DFT functional choice. For the explicit solvent approach, we significantly reduce the computational costs by embedding the microsolvated cluster in implicit bulk solvent, obtaining converged redox potential results with a smaller solvation shell. This combined implicit-explicit solvent model, together with GPU-accelerated quantum chemistry methods, enabled rapid generation of a large dataset of explicit-solvent-calculated redox potentials for 165 organic compounds, allowing detailed investigation of the error sources in explicit solvent redox potential calculations.

scholarly journals Bridging the experiment-calculation divide: machine learning corrections to redox potential calculations in implicit and explicit solvent models

2021 ◽  
Author(s):  
Eugen Hruska ◽  
Ariel Gale ◽  
Fang Liu
Keyword(s):  
Machine Learning ◽  
Redox Potential ◽  
Density Functional ◽  
Implicit Solvent ◽  
Systematic Bias ◽  
Redox Potentials ◽  
Solvent Models ◽  
Explicit Solvent ◽  
Solvent Model ◽  
Implicit And Explicit

Prediction of redox potentials is essential for catalysis and energy storage. Although density functional theory (DFT) calculations have enabled rapid redox potential predictions for numerous compounds, prominent errors persist compared to experimental measurements. In this work, we develop machine learning (ML) models to reduce the errors of redox potential calculations in both implicit and explicit solvent models. Training and testing of the ML correction models are based on the diverse ROP313 dataset with experimental redox potentials measured for organic and organometallic compounds in a variety of solvents. For the implicit solvent approach, our ML models can reduce both the systematic bias and the number of outliers. ML corrected redox potentials also demonstrate less sensitivity to DFT functional choice. For the explicit solvent approach, we significantly reduce the computational costs by embedding the microsolvated cluster in implicit bulk solvent, obtaining converged redox potential results with a smaller solvation shell. This combined implicit-explicit solvent model, together with GPU-accelerated quantum chemistry methods, enabled rapid generation of a large dataset of explicit-solvent-calculated redox potentials for 165 organic compounds, allowing detailed investigation of the error sources in explicit solvent redox potential calculations.

scholarly journals A Density Functional Theory-Based Scheme to Compute the Redox Potential of a Transition Metal Complex: Applications to Heme Compound

Molecules ◽  
2019 ◽  
Vol 24 (4) ◽  
pp. 819
Author(s):  
Toru Matsui ◽  
Jong-Won Song
Keyword(s):  
Dielectric Constant ◽  
Redox Potential ◽  
Density Functional ◽  
Absolute Error ◽  
Low Dielectric Constant ◽  
Density Functionals ◽  
Protonation State ◽  
Functional Theory ◽  
Low Dielectric ◽  
Solvent Model

We estimated the redox potential of a model heme compound by using the combination of our density functionals with a computational scheme, which corrects the solvation energy to the normal solvent model. Among many density functionals, the LC-BOP12 functional gave the smallest mean absolute error of 0.16 V in the test molecular sets. The application of these methods revealed that the redox potential of a model heme can be controlled within 200 mV by changing the protonation state and even within 20 mV by the flipping of the ligand histidine. In addition, the redox potential depends on the inverse of the dielectric constant, which controls the surroundings. The computational results also imply that a system with a low dielectric constant avoids the charged molecule by controlling either the redox potential or the protonation system.

Thermodynamics of Helix formation in small peptides of varying lengthin vacuo, implicit solvent and explicit solvent: Comparison between AMBER force fields

2019 ◽  
Vol 18 (03) ◽  
pp. 1950015
Author(s):  
Zhaoxi Sun ◽  
Xiaohui Wang
Keyword(s):  
Amino Acids ◽  
Free Energy ◽  
Hydrogen Bonds ◽  
Force Fields ◽  
Implicit Solvent ◽  
Conformational Ensemble ◽  
Helix Formation ◽  
Solvent Models ◽  
Explicit Solvent ◽  
Amber Force Fields

Helix formation is of great significance in protein folding. The helix-forming tendencies of amino acids are accumulated along the sequence to determine the helix-forming tendency of peptides. Computer simulation can be used to model this process in atomic details and give structural insights. In the current work, we employ equilibrate-state free energy simulation to systematically study the folding/unfolding thermodynamics of a series of mutated peptides. Two AMBER force fields including AMBER99SB and AMBER14SB are compared. The new 14SB force field uses refitted torsion parameters compared with 99SB and they share the same atomic charge scheme. We find that in vacuo the helix formation is mutation dependent, which reflects the different helix propensities of different amino acids. In general, there are helix formers, helix indifferent groups and helix breakers. The helical structure becomes more favored when the length of the sequence becomes longer, which arises from the formation of additional backbone hydrogen bonds in the lengthened sequence. Therefore, the helix indifferent groups and helix breakers will become helix formers in long sequences. Also, protonation-dependent helix formation is observed for ionizable groups. In 14SB, the helical structures are more stable than in 99SB and differences can be observed in their grouping schemes, especially in the helix indifferent group. In solvents, all mutations are helix indifferent due to protein–solvent interactions. The decrease in the number of backbone hydrogen bonds is the same with the increase in the number of protein–water hydrogen bonds. The 14SB in explicit solvent is able to capture the free energy minima in the helical state while 14SB in implicit solvent, 99SB in explicit solvent and 99SB in implicit solvent cannot. The helix propensities calculated under 14SB agree with the corresponding experimental values, while the 99SB results obviously deviate from the references. Hence, implicit solvent models are unable to correctly describe the thermodynamics even for the simple helix formation in isolated peptides. Well-developed force fields and explicit solvents are needed to correctly describe the protein dynamics. Aside from the free energy, differences in conformational ensemble under different force fields in different solvent models are observed. The numbers of hydrogen bonds formed under different force fields agree and they are mostly determined by the solvent model.

scholarly journals Saving Significant Amount of Time in MD Simulations by Using an Implicit Solvent Model and Elevated Temperatures

2013 ◽  
Vol 2013 ◽  
pp. 1-5 ◽  
Author(s):  
Ifat Shub ◽  
Ehud Schreiber ◽  
Yossef Kliger
Keyword(s):  
Elevated Temperatures ◽  
Md Simulations ◽  
Implicit Solvent ◽  
Dynamic Simulations ◽  
Implicit Solvent Model ◽  
Solvent Models ◽  
Solvent Model ◽  
Wide Range ◽  
Implicit Solvent Models ◽  
Computer Power

Molecular dynamic simulations are used for investigating various aspects of biological processes. Such simulations often require intensive computer power; therefore several solutions were developed to minimize the computer power needed, including the usage of elevated temperatures. Yet, such simulations are still not commonly used by the wide scientific community of chemists and biochemists. For about two years now, the molecular simulations suite GROMACS enables conducting simulations using implicit solvent models to further decrease runtimes. In order to quantify the saving in computer power, and to confirm the validity of the models, we followed the simple dissolution process of a single NaCl molecule. The results reveal approximately 350-fold decrease in real-world runtime when using an implicit solvent model and an elevated temperature, compared to using explicit water molecules and simulating at room temperature. In addition, in a wide range of temperatures, the dissolution times of NaCl are distributed, as expected, exponentially, both in explicit and in implicit solvent models, hence confirming the validity of the simulation approach. Hopefully, our findings will encourage many scientists to take advantage of the recent progress in the molecular dynamics field for various applications.

scholarly journals The Role of Displacing Confined Solvent in the Conformational Equilibrium of β-Cyclodextrin

2019 ◽  
Author(s):  
Peng He ◽  
Sheila Sarkar ◽  
Emilio Gallicchio ◽  
Tom Kurtzman ◽  
Lauren Wickstrom
Keyword(s):  
Conformational Equilibrium ◽  
Implicit Solvent ◽  
Theory Analysis ◽  
State Behavior ◽  
Generalized Born ◽  
Solvent Models ◽  
Explicit Solvent ◽  
Dynamics Simulations ◽  
Opening And Closing

<p>This study investigates the role of hydration and its relationship to the conformational equilibrium of the host molecule β-cyclodextrin. Molecular dynamics simulations indicate that the unbound β-cyclodextrin exhibits two state behavior in explicit solvent due to the opening and closing of its cavity. In implicit solvent, these transitions are not observed and there is one dominant conformation of β-cyclodextrin with an open cavity. Based on these observations, we investigate the hypothesis that the expulsion of thermodynamically unfavorable water molecules into the bulk plays an important role in controlling the accessibility of the closed macrostate at room temperature. We compare the results of the molecular mechanics analytical generalized Born plus non-polar solvation approach to those obtained through Grid Inhomogeneous Solvation Theory analysis with explicit solvation to elucidate the thermodynamic forces at play. The calculations help to illustrate the deficiencies of continuum solvent models and demonstrate the key role of the thermodynamics of enclosed hydration in driving the conformational equilibrium of molecules in solution. </p>

Sign in / Sign up

Export Citation Format

Share Document