Computer-assisted reaction prediction and synthesis design

The design of organic syntheses requires deep insight into chemical reactivity. Methods have been developed to calculate important electronic and energy effects and use them for the modelling of reaction mechanisms. The approach is illustrated by the haloform reaction. The procedures have been incorporated into different versions of the EROS system. A study of the synthesis of fredericamycin illustrates the use of these EROS procedures in synthesis design and reaction prediction. Recent work on synthesis design for aromatic compounds is highlighted and new definitions of the similarity of chemical compounds are discussed.     

A knowledge-based system for the automatic chronopotentiometric elucidation of electrode reaction mechanisms

A knowledge-based system for the elucidation of electrode reaction mechanisms based on chronopotentiometric experiments is described. The system runs the diagnostic experiments and uses the results in the reasoning process. New mechanistic knowledge can be added directly to its knowledge base in the form of production rules. The system is fully modular and its domain- specific modules can easily be changed for application to other electrochemical techniques. Correct operation of the system is demonstrated with the familiar reduction mechanisms of cadmium (II), zinc (II), cystamine and cinnamaldehyde.     

化学反应处理的计算模型

介绍了一种将同类反应上升为合成反应知识和在计算机上实现反合成分析的方法,反合成分析是合成设计中最关键的一步,在本工作中采用了基于谋略键寻找的合成设计方法。它有逻辑宜于在计算机上实现的优点。为了实现这个方法,我们首次提出了一种能中肯地描述合成反应的计算模型—反应知识的分类模型。这一模型由三条规则定义：规则A-反应类型;规则B-发生反应的外部条件;规则C-不适宜采用这个反应的情况;这种计算模型能够将海量反应数据中最重要最基本的信息提炼出来,转换成计算机能处理的知识。它也包含有反应适用范围的信息,从而提高了析分过程的外推能力。     

RCARM: Reaction classification using automated reaction mapping

Detailed chemical kinetic modeling of gas‐phase reactions can result in automatically generated mechanisms that contain thousands of reactions. In this paper, we describe the development of a rule‐based expert system tool that organizes these reactions into classes such as hydrogen abstraction and beta scission. We have developed 29 simple classification rules, 20 complex (well‐skipping) classification rules, and four second‐stage classification rules. This greatly simplifies the task of the chemical kineticist who wishes to verify, analyze, and gain insights into the reactions comprising the mechanism. This system, which is based on the automated identification of the bonds that break and form in a chemical reaction (the reaction mapping problem), is used to classify reactions in three different mechanisms. © 2012 Wiley Periodicals, Inc. Int J Chem Kinet 45: 125–139, 2013     

Synergy Between Expert and Machine-Learning Approaches Allows for Improved Retrosynthetic Planning

When computers plan multistep syntheses, they can rely either on expert knowledge or information machine-extracted from large reaction repositories. Both approaches suffer from imperfect functions evaluating reaction choices: expert functions are heuristics based on chemical intuition, whereas machine learning (ML) relies on neural networks (NNs) that can make meaningful predictions only about popular reaction types. This paper shows that expert and ML approaches can be synergistic—specifically, when NNs are trained on literature data matched onto high-quality, expert-coded reaction rules, they achieve higher synthetic accuracy than either of the methods alone and, importantly, can also handle rare/specialized reaction types.     

Synergy Between Expert and Machine‐Learning Approaches Allows for Improved Retrosynthetic Planning

When computers plan multistep syntheses, they can rely either on expert knowledge or information machine‐extracted from large reaction repositories. Both approaches suffer from imperfect functions evaluating reaction choices: expert functions are heuristics based on chemical intuition, whereas machine learning (ML) relies on neural networks (NNs) that can make meaningful predictions only about popular reaction types. This paper shows that expert and ML approaches can be synergistic—specifically, when NNs are trained on literature data matched onto high‐quality, expert‐coded reaction rules, they achieve higher synthetic accuracy than either of the methods alone and, importantly, can also handle rare/specialized reaction types.     

Activation energy for a model ferrous-ferric half reaction from transition path sampling

Activation parameters for the model oxidation half reaction of the classical aqueous ferrous ion are compared for different molecular simulation techniques. In particular, activation free energies are obtained from umbrella integration and Marcus theory based thermodynamic integration, which rely on the diabatic gap as the reaction coordinate. The latter method also assumes linear response, and both methods obtain the activation entropy and the activation energy from the temperature dependence of the activation free energy. In contrast, transition path sampling does not require knowledge of the reaction coordinate and directly yields the activation energy [C. Dellago and P. G. Bolhuis, Mol. Simul. 30, 795 (2004)]. Benchmark activation energies from transition path sampling agree within statistical uncertainty with activation energies obtained from standard techniques requiring knowledge of the reaction coordinate. In addition, it is found that the activation energy for this model system is significantly smaller than the activation free energy for the Marcus model, approximately half the value, implying an equally large entropy contribution.     

Comparison of reaction pathways calculated by different algorithms for disilane and water trimer

There still exists some confusion in the literature concerning the definition of a minimum energy pathway and the coordinate system in which it is calculated. Here we compare steepest-descent and eigenvector-following pathways, both with and without a mass-weighted metric. The systems studied are disilane and the water trimer, and we employ various basis sets at the SCF level of theory. We find that paths calculated using eigenvector-following and steepestdescent are practically the same, at least in terms of the reaction mechanism.We find that for the mass-weighted metric the pathways are similar, although in principle they do not have to be identical. Finally, we verify that the geometrical symmetry selection rules hold for a pathway mediated by a recently discovered transition state of the disilane system.     

Spectrophotometric reaction rate method for determination of barbituric acid by inhibition of the hydrochloric acid-bromate reaction

A new kinetic-spectrophotometric method was developed for the determination of barbituric acid. The method is based on its inhibition effect on the reaction between hydrochloric acid and bromate. The decolorization of methyl orange by the reaction products was used to monitor the reaction spectrophotometrically at 510 nm. The variable affecting the rate of the reaction was investigated. The method is simple, rapid, relatively sensitive and precise. The limit of detection is 7.9x10(-7) M and calibration rang is 1x10(-6)-6.0x10(-4) M barbituric acid. The linearity range of the calibration graph is depends on bromate concentration. The relative standard deviation of seven replication determination of 5.6x10(-6) M barbituric acid was 1.8%. The influence of potential interfering substance was studied.     

Superstructure searching algorithm for generic reaction retrieval

Chemical reaction knowledge is usually summarized and retrieved by chemists from references, journals, and reaction databases. To rigorously extract chemical reaction knowledge from large data sets, computer algorithms become much more important. This paper presents a new approach, superstructure searching (SSS) algorithm, for generic reaction retrieval. The algorithm considers all known reaction patterns from the targeted structure and assigns synthetic routes for new chemical compounds. This algorithm consists of screening, atom-by-atom comparison, and computation of R-groups' similarity.     

Rank annihilation factor analysis method for spectrophotometric study of second-order reaction kinetics

In this study several methods are described to determine the rate constant of a second-order reaction in the form of A+B→C. These approaches allow circumventing a rank deficiency inherent of a second-order reaction when the spectroscopic data is influenced by additional source of variance. Classically, to determine the unknown rate constant in this kind of systems, one needs to have extra knowledge about the system, including the spectra of the reactants or product and the exact kinetics. In the case of the presence of an unknown phenomenon in the data set that cannot be explained by the model, such as baseline drift, the estimated rate constant might be erroneous. Present work is a modification of the rank annihilation factor analysis (RAFA) algorithm by inclusion of I) pure spectra of reactants, or IIA) mean centering step, or IIB) mean spectrum. The proposed methods can interestingly be applied on a single kinetic run. The performances of the new methods have been evaluated by applying them to analysis of simulated and experimental data.     

An inverse problem in reaction kinetics

In this paper we investigate the problem of extracting information about chemical reactions involving multiple species from the time history of the concentration of each species. The mathematical model of the kinetic system leads to a system of ordinary differential equations. Our focus is to examine whether the species’ concentrations as functions of time are sufficient to determine what chemical reactions, and at what reaction rates, have occurred. We show that within the limitation of our model, there may be many candidate reaction systems that could explain the data. Using the notion of sparsity, we provide a quantitative assessment of the question of distinguishability. We further demonstrate that sparsity enforcing approaches, such as minimizing the ℓ ₁ or the ℓ ₀ norms are not reliable. Our conclusion is that additional knowledge about the kinetic system will be necessary to reliably solve this inverse problem.     

Efficiency analysis of reaction rate calculation methods using analytical models I: The two-dimensional sharp barrier

We analyze the efficiency of different methods for the calculation of reaction rates in the case of a simple two-dimensional analytical benchmark system. Two classes of methods are considered: the first is based on the free energy calculation along a reaction coordinate and the calculation of the transmission coefficient, the second on the sampling of dynamical pathways. We give scaling rules for how this efficiency depends on barrier height and width, and we hand out simple optimization rules for the method-specific parameters. We show that the path sampling methods, using the transition interface sampling technique, become exceedingly more efficient than the others when the reaction coordinate is not the optimal one.     

Understanding the physics and chemistry of reaction mechanisms from atomic contributions: a reaction force perspective

Studying chemical reactions involves the knowledge of the reaction mechanism. Despite activation barriers describing the kinetics or reaction energies reflecting thermodynamic aspects, identifying the underlying physics and chemistry along the reaction path contributes essentially to the overall understanding of reaction mechanisms, especially for catalysis. In the past years the reaction force has evolved as a valuable tool to discern between structural changes and electrons' rearrangement in chemical reactions. It provides a framework to analyze chemical reactions and additionally a rational partition of activation and reaction energies. Here, we propose to separate these energies further in atomic contributions, which will shed new insights in the underlying reaction mechanism. As first case studies we analyze two intramolecular proton transfer reactions. Despite the atom based separation of activation barriers and reaction energies, we also assign the participation of each atom in structural changes or electrons' rearrangement along the intrinsic reaction coordinate. These participations allow us to identify the role of each atom in the two reactions and therfore the underlying chemistry. The knowledge of the reaction chemistry immediately leads us to suggest replacements with other atom types that would facilitate certain processes in the reaction. The characterization of the contribution of each atom to the reaction energetics, additionally, identifies the reactive center of a molecular system that unites the main atoms contributing to the potential energy change along the reaction path.     

Off-line form of the Michaelis-Menten equation for studying the reaction kinetics in a polymer microchip integrated with enzyme microreactor

We firstly transformed the traditional Michaelis-Menten equation into an off-line form which can be used for evaluating the Michaelis-Menten constant after the enzymatic reaction. For experimental estimation of the kinetics of enzymatic reactions, we have developed a facile and effective method by integrating an enzyme microreactor into direct-printing polymer microchips. Strong nonspecific adsorption of proteins was utilized to effectively immobilize enzymes onto the microchannel wall, forming the integrated on-column enzyme microreactor in a microchip. The properties of the integrated enzyme microreactor were evaluated by using the enzymatic reaction of glucose oxidase (GOx) with its substrate glucose as a model system. The reaction product, hydrogen peroxide, was electrochemically (EC) analyzed using a Pt microelectrode. The data for enzyme kinetics using our off-line form of the Michaelis-Menten equation was obtained (K(m) = 2.64 mM), which is much smaller than that reported in solution (K(m) = 6.0 mM). Due to the hydrophobic property and the native mesoscopic structure of the poly(ethylene terephthalate) film, the immobilized enzyme in the microreactor shows good stability and bioactivity under the flowing conditions.     

A quantum theory of chemical processes and reaction rates based on diabatic electronic functions coupled in an external field

The method discussed in this work provides a theoretical framework where simple chemical reactions resemble any other standard quantum process, i.e., a transition in quantum state mediated by the electromagnetic field. In our approach, quantum states are represented as a superposition of electronic diabatic basis functions, whose amplitudes can be modulated by the field and by the external control of nuclear configurations. Using a one-dimensional three-state model system, we show how chemical structure and dynamics can be represented in terms of these control parameters, and propose an algorithm to compute the reaction probabilities. Our analysis of effective energy barriers generalizes previous ideas on structural similarity between reactant, and product, and transition states using the geometry of conventional reaction paths. In the present context, exceptions to empirical rules such as the Hammond postulate appear as effects induced by the environment that supplies the external field acting on the quantum system.     

Reduction of chemical reaction networks using quasi-integrals

We present a numerical method to identify possible candidates for quasi-stationary manifolds in complex reaction networks governed by systems of ordinary differential equations. Inspired by singular perturbation theory, we examine the ratios of certain components of the reaction rate vector. Those ratios that rapidly approach a nearly constant value define a slow manifold for the original flow in terms of quasi-integrals, that is, functions that are nearly constant along the trajectories. The dimensionality of the original system is thus effectively reduced without reliance on a priori knowledge of the different time scales in the system. We also demonstrate the relation of our approach to singular perturbation theory which, in its simplest form, is just the well-known quasi-steady-state approximation. In two case studies, we apply our method to oscillatory chemical systems: the 6-dimensional hemin-hydrogen peroxide-sulfite pH oscillator and a 10-dimensional mechanistic model for the peroxidase-oxidase (PO) reaction system. We conjecture that the presented method is especially suited for a straightforward reduction of higher dimensional dynamical systems where analytical methods fail to identify the different time scales associated with the slow invariant manifolds present in the system.