Automated Preparation of Nanoscopic Structures: Graph-Based Sequence Analysis, Mismatch Detection, and pH-Consistent Protonation with Uncertainty Estimates DOI Creative Commons
Katja‐Sophia Csizi, Markus Reiher

arXiv (Cornell University), Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 1, 2023

Structure and function in nanoscale atomistic assemblies are tightly coupled, every atom with its specific position even electron will have a decisive effect on the electronic structure, hence, molecular properties. Molecular simulations of nanoscopic structures therefore require accurately resolved three-dimensional input structures. If extracted from experiment, these often suffer severe uncertainties, which lack information hydrogen atoms is prominent example. Hence, experimental careful review curation, time-consuming error-prone process. Here, we present fast robust protocol for automated structure analysis, pH-consistent protonation, short, ASAP. For biomolecules as target, ASAP integrates sequence analysis error assessment given structure. allows pKa prediction reference data through Gaussian process regression including uncertainty estimation connects to system-focused modeling described (J. Chem. Theory Comput. 16, 2020, 1646). Although focused biomolecules, can be extended other objects, because most design elements rely general graph-based foundation guaranteeing transferability. The modular character underlying pipeline supports different degrees automation, (i) efficient feedback loops human-machine interaction low entrance barrier (ii) integration into autonomous procedures such force field parametrizations. This facilitates switching pH-state on-the-fly reparametrization during simulation at virtually no extra computational cost.

Language: Английский

Exploration of Reaction Pathways and Chemical Transformation Networks DOI Creative Commons
Gregor N. C. Simm, Alain C. Vaucher, Markus Reiher

et al.

The Journal of Physical Chemistry A, Journal Year: 2018, Volume and Issue: 123(2), P. 385 - 399

Published: Nov. 13, 2018

For the investigation of chemical reaction networks, identification all relevant intermediates and elementary reactions is mandatory. Many algorithmic approaches exist that perform explorations efficiently in an automated fashion. These differ their application range, level completeness exploration, amount heuristics human intervention required. Here, we describe compare different based on these criteria. Future directions leveraging strengths heuristics, interaction, physical rigor are discussed.

Language: Английский

Citations

218
Chemoton 2.0: Autonomous Exploration of Chemical Reaction Networks DOI Creative Commons
Jan P. Unsleber, Stephanie A. Grimmel, Markus Reiher

et al.

Journal of Chemical Theory and Computation, Journal Year: 2022, Volume and Issue: 18(9), P. 5393 - 5409

Published: Aug. 4, 2022

Fueled by advances in hardware and algorithm design, large-scale automated explorations of chemical reaction space have become possible. Here, we present our approach to an open-source, extensible framework for mechanisms based on the first principles quantum mechanics. It is intended facilitate network diverse problems with a wide range goals such as mechanism elucidation, path optimization, retrosynthetic validation, reagent microkinetic modeling. The stringent first-principles basis all algorithms key general applicability that avoids any restrictions specific systems. Such agile requires multiple specialized software components which three modules this work. module, Chemoton,drives exploration networks. For itself, introduce two new elementary-step searches are Newton trajectories. performance these assessed variety reactions characterized broad diversity terms bonding patterns elements. We reproduce significantly extend what known about provide resulting data be used starting point further future reference.

Language: Английский

Citations

56
Mechanism Deduction from Noisy Chemical Reaction Networks DOI
Jonny Proppe, Markus Reiher

Journal of Chemical Theory and Computation, Journal Year: 2018, Volume and Issue: 15(1), P. 357 - 370

Published: Dec. 3, 2018

We introduce KiNetX, a fully automated meta-algorithm for the kinetic analysis of complex chemical reaction networks derived from semiaccurate but efficient electronic structure calculations. It is designed to (i) accelerate exploration such and (ii) cope with model-inherent errors in calculations on elementary steps. developed implemented KiNetX possess three features. First, evaluates relevance every species (yet incomplete) network confine search new steps only those that are considered possibly relevant. Second, identifies eliminates all kinetically irrelevant reactions reduce graph comprehensible mechanism. Third, estimates sensitivity concentrations toward changes individual rate constants (derived relative free energies), which allows us systematically select most model each given predefined accuracy. The novelty consists rigorous propagation correlated free-energy uncertainty through our analyis. To examine performance we AutoNetGen. semirandomly generates chemistry-mimicking by encoding logic into their underlying structure. AutoNetGen consider vast number distinct chemistry-like scenarios and, hence, discuss importance statistical context. Our results reveal reliably supports deduction product ratios, dominant pathways, other properties data.

Language: Английский

Citations

57
Automated Construction of Quantum–Classical Hybrid Models DOI Creative Commons
Christoph Brunken, Markus Reiher

Journal of Chemical Theory and Computation, Journal Year: 2021, Volume and Issue: 17(6), P. 3797 - 3813

Published: May 12, 2021

We present a protocol for the fully automated construction of quantum mechanical-(QM)-classical hybrid models by extending our previously reported approach on self-parametrizing system-focused atomistic (SFAM) J. Chem. Theory Comput. 2020, 16, 1646]. In this QM/SFAM approach, size and composition QM region is evaluated in an manner based first principles so that model describes atomic forces center accurately. This entails evaluation differently sized regions with bearable computational overhead needs to be paid validation procedures. Applying SFAM classical part eliminates any dependence pre-existing parameters due its mechanically derived parametrization. Hence, capable delivering high fidelity complete automation. Furthermore, since are generated whole system, ansatz allows convenient re-definition during molecular exploration. For purpose, local re-parametrization scheme introduced, which efficiently generates additional fly when new covalent bonds formed (or broken) moved region.

Language: Английский

Citations

36
Self-Parametrizing System-Focused Atomistic Models DOI
Christoph Brunken, Markus Reiher

Journal of Chemical Theory and Computation, Journal Year: 2020, Volume and Issue: 16(3), P. 1646 - 1665

Published: Jan. 17, 2020

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

Language: Английский

Citations

37
Computational Tools for the Prediction of Site- and Regioselectivity of Organic Reactions DOI Creative Commons
Lukas M. Sigmund,

Michele Assante,

Magnus J. Johansson

et al.

Chemical Science, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 1, 2025

This article reviews computational tools for the prediction of regio- and site-selectivity organic reactions. It spans from quantum chemical procedures to deep learning models showcases application presented tools.

Language: Английский

Citations

0
Systematic microsolvation approach with a cluster‐continuum scheme and conformational sampling DOI
Gregor N. C. Simm, Paul L. Türtscher, Markus Reiher

et al.

Journal of Computational Chemistry, Journal Year: 2020, Volume and Issue: 41(12), P. 1144 - 1155

Published: Feb. 6, 2020

Abstract Solvation is a notoriously difficult and nagging problem for the rigorous theoretical description of chemistry in liquid phase. Successes failures various approaches ranging from implicit solvation modeling through dielectric continuum embedding microsolvated quantum chemical to explicit molecular dynamics highlight this situation. Here, we focus on microsolvation discuss an conformational sampling ansatz make approach systematic. For purpose, introduce algorithm rolling automated solutes. Our protocol takes rearrangements solvent shell into account. Its reliability assessed by monitoring evolution spread average observables interest.

Language: Английский

Citations

31
Automated Preparation of Nanoscopic Structures: Graph-Based Sequence Analysis, Mismatch Detection, and pH-Consistent Protonation with Uncertainty Estimates DOI Creative Commons
Katja‐Sophia Csizi, Markus Reiher

arXiv (Cornell University), Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 1, 2023

Structure and function in nanoscale atomistic assemblies are tightly coupled, every atom with its specific position even electron will have a decisive effect on the electronic structure, hence, molecular properties. Molecular simulations of nanoscopic structures therefore require accurately resolved three-dimensional input structures. If extracted from experiment, these often suffer severe uncertainties, which lack information hydrogen atoms is prominent example. Hence, experimental careful review curation, time-consuming error-prone process. Here, we present fast robust protocol for automated structure analysis, pH-consistent protonation, short, ASAP. For biomolecules as target, ASAP integrates sequence analysis error assessment given structure. allows pKa prediction reference data through Gaussian process regression including uncertainty estimation connects to system-focused modeling described (J. Chem. Theory Comput. 16, 2020, 1646). Although focused biomolecules, can be extended other objects, because most design elements rely general graph-based foundation guaranteeing transferability. The modular character underlying pipeline supports different degrees automation, (i) efficient feedback loops human-machine interaction low entrance barrier (ii) integration into autonomous procedures such force field parametrizations. This facilitates switching pH-state on-the-fly reparametrization during simulation at virtually no extra computational cost.

Language: Английский

Citations

0