scholarly journals Machine learning meets mechanistic modelling for accurate prediction of experimental activation energies

2021 ◽  
Author(s):  
Kjell Jorner ◽  
Tore Brinck ◽  
Per-Ola Norrby ◽  
David Buttar
Keyword(s):  
Machine Learning ◽  
Activation Energies ◽  
Accurate Prediction ◽  
Nucleophilic Aromatic Substitution ◽  
Aromatic Substitution ◽  
Mechanistic Modelling

Hybrid reactivity models, combining mechanistic calculations and machine learning with descriptors, are used to predict barriers for nucleophilic aromatic substitution.

scholarly journals Machine Learning Meets Mechanistic Modelling for Accurate Prediction of Experimental Activation Energies

2020 ◽  
Author(s):  
Kjell Jorner ◽  
Tore Brinck ◽  
Per-Ola Norrby ◽  
David Buttar
Keyword(s):  
Machine Learning ◽  
Density Functional Theory ◽  
Transition State ◽  
Density Functional ◽  
Structural Information ◽  
Hybrid Models ◽  
Accurate Prediction ◽  
Nucleophilic Aromatic Substitution ◽  
Functional Theory ◽  
Promising Alternative

Accurate prediction of chemical reactions in solution is challenging for current state-of-the-art approaches based on transition state modelling with density functional theory. Models based on machine learning have emerged as a promising alternative to address these problems, but these models currently lack the precision to give crucial information on the magnitude of barrier heights, influence of solvents and catalysts and extent of regio- and chemoselectivity. Here, we construct hybrid models which combine the traditional transition state modelling and machine learning to accurately predict reaction barriers. We train a Gaussian Process Regression model to reproduce high-quality experimental kinetic data for the nucleophilic aromatic substitution reaction and use it to predict barriers with a mean absolute error of 0.77 kcal/mol for an external test set. The model was further validated on regio- and chemoselectivity prediction on patent reaction data and achieved a competitive top-1 accuracy of 86%, despite not being trained explicitly for this task. Importantly, the model gives error bars for its predictions that can be used for risk assessment by the end user. Hybrid models emerge as the preferred alternative for accurate reaction prediction in the very common low-data situation where only 100–150 rate constants are available for a reaction class. With recent advances in deep learning for quickly predicting barriers and transition state geometries from density functional theory, we envision that hybrid models will soon become a standard alternative to complement current machine learning approaches based on ground-state physical organic descriptors or structural information such as molecular graphs or fingerprints.

scholarly journals Machine Learning Meets Mechanistic Modelling for Accurate Prediction of Experimental Activation Energies

2020 ◽  
Author(s):  
Kjell Jorner ◽  
Tore Brinck ◽  
Per-Ola Norrby ◽  
David Buttar
Keyword(s):  
Machine Learning ◽  
Density Functional Theory ◽  
Transition State ◽  
Density Functional ◽  
Structural Information ◽  
Hybrid Models ◽  
Accurate Prediction ◽  
Nucleophilic Aromatic Substitution ◽  
Functional Theory ◽  
Promising Alternative

Accurate prediction of chemical reactions in solution is challenging for current state-of-the-art approaches based on transition state modelling with density functional theory. Models based on machine learning have emerged as a promising alternative to address these problems, but these models currently lack the precision to give crucial information on the magnitude of barrier heights, influence of solvents and catalysts and extent of regio- and chemoselectivity. Here, we construct hybrid models which combine the traditional transition state modelling and machine learning to accurately predict reaction barriers. We train a Gaussian Process Regression model to reproduce high-quality experimental kinetic data for the nucleophilic aromatic substitution reaction and use it to predict barriers with a mean absolute error of 0.77 kcal/mol for an external test set. The model was further validated on regio- and chemoselectivity prediction on patent reaction data and achieved a competitive top-1 accuracy of 86%, despite not being trained explicitly for this task. Importantly, the model gives error bars for its predictions that can be used for risk assessment by the end user. Hybrid models emerge as the preferred alternative for accurate reaction prediction in the very common low-data situation where only 100–150 rate constants are available for a reaction class. With recent advances in deep learning for quickly predicting barriers and transition state geometries from density functional theory, we envision that hybrid models will soon become a standard alternative to complement current machine learning approaches based on ground-state physical organic descriptors or structural information such as molecular graphs or fingerprints.

Enantioselective Nucleophilic Aromatic Substitution Reaction of Azlactones to Synthesize Quaternary α-Amino Acid Derivatives

Synlett ◽  
2020 ◽  
Author(s):  
Xiaohua Liu ◽  
Yi Li ◽  
Hao Pan ◽  
Wang-Yuren Li ◽  
Xiaoming Feng
Keyword(s):  
Hydrogen Bonding ◽  
Charge Transfer ◽  
Amino Acid ◽  
Substitution Reaction ◽  
Nucleophilic Aromatic Substitution ◽  
Amino Acid Derivatives ◽  
Amino Acid Esters ◽  
Aromatic Substitution ◽  
Charge Transfer Interaction ◽  
Acid Esters

AbstractAn asymmetric organocatalytic nucleophilic aromatic substitution reaction of azlactones with electron-deficient aryls was established. A variety of α-aryl α-alkyl α-amino acid esters and peptides were obtained in decent yields and stereoselectivities. A new bifunctional catalytic mode involving charge-transfer interaction and hydrogen bonding is proposed to explain the enantioselectivity.

scholarly journals From Black Box to Machine Learning: A Journey through Membrane Process Modelling

Membranes ◽  
2021 ◽  
Vol 11 (8) ◽  
pp. 574
Author(s):  
Claudia F. Galinha ◽  
João G. Crespo
Keyword(s):  
Machine Learning ◽  
Process Modelling ◽  
Model Membrane ◽  
Point Of View ◽  
Personal Perspective ◽  
Process Conditions ◽  
Monitoring And Control ◽  
Membrane Processes ◽  
Mechanistic Modelling ◽  
Process Monitoring And Control

Membrane processes are complex systems, often comprising several physicochemical phenomena, as well as biological reactions, depending on the systems studied. Therefore, process modelling is a requirement to simulate (and predict) process and membrane performance, to infer about optimal process conditions, to assess fouling development, and ultimately, for process monitoring and control. Despite the actual dissemination of terms such as Machine Learning, the use of such computational tools to model membrane processes was regarded by many in the past as not useful from a scientific point-of-view, not contributing to the understanding of the phenomena involved. Despite the controversy, in the last 25 years, data driven, non-mechanistic modelling is being applied to describe different membrane processes and in the development of new modelling and monitoring approaches. Thus, this work aims at providing a personal perspective of the use of non-mechanistic modelling in membrane processes, reviewing the evolution supported in our own experience, gained as research group working in the field of membrane processes. Additionally, some guidelines are provided for the application of advanced mathematical tools to model membrane processes.

Nucleophilic aromatic substitution using Et3SiH/cat. t-Bu-P4 as a system for nucleophile activation

2007 ◽  
pp. 2264 ◽  
Author(s):  
Masahiro Ueno ◽  
Misato Yonemoto ◽  
Masahiro Hashimoto ◽  
Andrew E. H. Wheatley ◽  
Hiroshi Naka ◽  
...  
Keyword(s):  
Nucleophilic Aromatic Substitution ◽  
Aromatic Substitution ◽  
Nucleophile Activation
Sign in / Sign up

Export Citation Format

Share Document