Analytical hessian fitting schemes for efficient determination of force‐constant parameters in molecular mechanics

Wang, Ruixing; Ozhgibesov, Mikhail; Hirao, Hajime

doi:10.1002/jcc.25100

Cited by 16 publications

(18 citation statements)

References 63 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…99,100 A small set of 11 small and medium-sized organic and inorganic molecules was prepared and the detailed results are provided in the Supporting Information. The mean absolute error (MAE) of the wavenumbers calculated by SFAM compared to their reference for the molecules in this set range between 12.7 and 116.6 cm −1 , which is in line with the results obtained by Hirao and coworkers in their study 85 of the partial Hessian fitting approach. The correlation between the SFAM frequencies and their reference values is presented in Fig.…”

Section: Ii5 Accuracy Of the MM Modelsupporting

confidence: 90%

Self-Parametrizing System-Focused Atomistic Models

Brunken

Reiher

2020

J. Chem. Theory Comput.

View full text Add to dashboard Cite

Computational studies of chemical reactions in complex environments such as proteins, nanostructures, or on surfaces require accurate and efficient atomistic models applicable to the nanometer scale. In general, an accurate parametrization of the atomistic entities will not be available for arbitrary system classes, but demands a fast automated system-focused parametrization procedure to be quickly applicable, reliable, flexible, and reproducible. Here, we develop and combine an automatically parametrizable quantum chemically derived molecular mechanics model with machine-learned corrections under autonomous uncertainty quantification and refinement. Our approach first generates an accurate, physically motivated model from a minimum energy structure and its corresponding Hessian matrix by a partial Hessian fitting procedure of the force constants. This model is then the starting point to generate a large number of configurations for which additional off-minimum reference data can be evaluated on the fly. A ∆-machine learning model is trained on these data to provide a correction to energies and forces including uncertainty estimates. During the procedure, the flexibility of the machine learning model is tailored to the amount of available training data. The parametrization of large systems is enabled by a fragmentation approach. Due to their modular nature, all model construction steps allow for model improvement in a rolling fashion. Our approach may also be employed for the generation of system-focused electrostatic molecular mechanics embedding environments in a quantum mechanical molecular-mechanical hybrid model for arbitrary atomistic structures at the nanoscale.

show abstract

Section: Ii5 Accuracy Of the MM Modelsupporting

confidence: 90%

Self-Parametrizing System-Focused Atomistic Models

Brunken

Reiher

2020

J. Chem. Theory Comput.

View full text Add to dashboard Cite

show abstract

“…For system minimisations and MD simulations the Amber ff14SB force field was used [31, 32] . The missing parameters for the active site of the protein were calculated based on a B3LYP‐D3/def2‐SVP‐optimised model of Fe II and its first shell ligands (His‐144, Glu‐146, His‐279, water molecule and truncated 2OG) by using the PARMHESS program with the internal Hessian fitting (IHF) scheme [33, 34] . The same procedure was used for the oxidised iron cofactor using the model of Fe IV –oxo and its ligands (His‐144, Glu‐146, His‐279 and truncated succinate).…”

Section: Methodsmentioning

confidence: 99%

Enzyme Multifunctionality by Control of Substrate Positioning Within the Catalytic Cycle—A QM/MM Study of Clavaminic Acid Synthase

Wojdyla

Borowski

2020

Chemistry A European J

View full text Add to dashboard Cite

Clavaminic acid synthase from Streptomyces clavuligerus is an FeII/2‐oxoglutarate‐dependent dioxygenase, crucial for the biosynthesis of the β‐lactamase inhibitor clavulanic acid. It catalyses three consecutive oxidative reactions, that is, hydroxylation, cyclisation and desaturation, in a single binding cavity. As follows from the results of this QM/MM study, CAS versatility and selectivity depends on the binding cavity, which interplays differently with the substrate for each reaction. The enzyme–substrate interactions affect the substrate's ability to re‐position during the reaction, either constraining it in its primary position, which impedes processes other than oxygen rebound, or allowing change, which facilitates desaturation. This differential effect originates from two aspartate residues, which strongly interact with the guanidine group of the hydroxylation substrate and stabilise the orientation of the molecule. These residues interact less effectively with the smaller amine group of the desaturation substrate(s), aiding their re‐positioning and the subsequent formation of a double bond.

show abstract

“…Hence, if one uses within the amber force field bond and angle force constants derived with the Seminario method together with atomic partial charges and atomic vdW (Lennard-Jones) parameters, a potential problem of doublecounting of non-bonded interactions arises. [4,5] . Here, we have used ParmHess to derive force constants for bonds and angles involving Fe.…”

Section: Parmhessmentioning

confidence: 99%

“…Here, we are presenting our efforts to find the best set of force field parameters to describe the geometry of the Fe(II) cofactor present in the active site of ectoine synthase (EctC). Experimental data strongly indicate that the Fe(II) ion plays a central role in catalytic activity of EctC [1,2], thus we aim to obtain as precise description of the metal site as possible by testing three parameterization methods for bonded approach: the projected Hessian method proposed by Seminario [3] , Hessian-based methods developed and implemented within the ParmHess program by R. Wang et al [4,5] and the energybased method implemented within the ParamFit program by Betz and Walker [6] . Over the years, thorough surveys of the metal site parameterization methods have been published.…”

Section: Introductionmentioning

confidence: 99%

Comparison of different approaches to derive classical bonded force-field parameters for a transition metal cofactor: a case study for non-heme iron site of ectoine synthase

2021

View full text Add to dashboard Cite

Computational investigations into the structure and function of metalloenzymes with transition metal cofactors require proper preparation of the model, which requires obtaining reliable force field parameters for the cofactor. Here, we present a test case where several methods were used to derive amber force field parameters for a bonded model of the Fe(II) cofactor of ectoine synthase. Moreover, the spin of the ground state of the cofactor was probed by DFT and post-HF methods, which consistently indicated the quintet state is lowest in energy and well separated from triplet and singlet. The performance of the obtained force field parameter sets, derived for the quintet spin state, was scrutinized and compared taking into account metrics focused on geometric features of the models as well as their energetics. The main conclusion of this study is that Hessian-based methods yield parameters which represent the geometry around the metal ion, but poorly reproduce energy variance with geometrical changes. On the other hand, the energy-based method yields parameters accurately reproducing energy-structure relationships, but with bad performance in geometry optimization. Preliminary tests show that admixing geometrical criteria to energy-based methods may allow to derive parameters with acceptable performance for both energy and geometry.

show abstract

Analytical hessian fitting schemes for efficient determination of force‐constant parameters in molecular mechanics

Cited by 16 publications

References 63 publications

Self-Parametrizing System-Focused Atomistic Models

Self-Parametrizing System-Focused Atomistic Models

Enzyme Multifunctionality by Control of Substrate Positioning Within the Catalytic Cycle—A QM/MM Study of Clavaminic Acid Synthase

Comparison of different approaches to derive classical bonded force-field parameters for a transition metal cofactor: a case study for non-heme iron site of ectoine synthase

Contact Info

Product

Resources

About