Reaction-based machine learning representations for predicting the enantioselectivity of organocatalysts

Hundreds of catalytic methods are developed each year to meet the demand for high-purity chiral compounds. The computational design of enantioselective organocatalysts remains a significant challenge, as catalysts are typically discovered through experimental screening. Recent advances in combining quantum chemical computations and machine learning (ML) hold great potential to propel the next leap forward in asymmetric catalysis. Within the context of quantum chemical machine learning (QML, or atomistic ML), the ML representations used to encode the three-dimensional structure of molecules and evaluate their similarity cannot easily capture the subtle energy differences that govern enantioselectivity. Here, we present a general strategy for improving molecular representations within an atomistic machine learning model to predict the DFT-computed enantiomeric excess of asymmetric propargylation organocatalysts solely from the structure of catalytic cycle intermediates. Mean absolute errors as low as 0.25 kcal mol⁻¹ were achieved in predictions of the activation energy with respect to DFT computations. By virtue of its design, this strategy is generalisable to other ML models, to experimental data and to any catalytic asymmetric reaction, enabling the rapid screening of structurally diverse organocatalysts from available structural information.

A machine learning model for enantioselectivity prediction using reaction-based molecular representations.     

Diffusion-controlled vinyl polymerization. I. The gel effect

It is well known that the reaction rate and molecular weight of vinyl polymers can change markedly during the course of polymerization and that these changes are due to the influence of diffusion on the termination reaction. The chain length dependence of the termination rate constant has been considered in this work and has resulted in a general method of treating the polymerization kinetics and molecular weight distribution. This method is independent of the form of the chain length dependency and is capable of dealing with both disproportionation and recombination modes of termination. A specific model for the termination rate constant with chain length dependence is proposed and is based on free volume theory and entanglement coupling. Master curves for the characteristics of the reaction rate and molecular weight distribution are presented with the application of this model.     

Gitoxigenin glycosides

Summary Three cardiac glycosides have been synthesized from gitoxigenin and D-glucose: gitoxigenin 3-O--D-glucopyranoside, gitoxigenin 16 -O- -D-glucopyranoside, and gitoxigenin 3, 16-di-O- -D-glucoside. The first glycoside, obtained with a yield of 49% on the weight of the initial aglycone, possesses a fairly high biological activity (0.25 mg/kg), while the other two, formed in a yield of about 0.3%, do not possess the action on the heart that is specific for the cardenolides.Khar'kov Chemical and Pharmaceutical Scientific-Research Institute. Translated from Khimiya Prirodnykh Soedinenii, No. 1, pp. 45–49, January, 1971.     

Double-layer effects on simple electrode reactions III. The reduction of Eu3+ and Cr3+ in the presence of thiocyanate anions. Evidence for ligand bridging

The rates of reduction of Eu³⁺ and Cr³⁺ have been measured in mixed perchlorate + thiocyanate electrolytes at constant ionic strength, using low concentrations of thiocyanate to minimize its association with the cationic reactants. The effect of adsorbed thiocyanate anions on the reduction kinetics of Cr³⁺ resembled those produced by iodide and bromide on both Cr³⁺ and Eu³⁺. However, thiocyanate exhibited an unusual catalytic effect on the reduction of Eu³⁺ which was identified as arising from a reaction pathway involving thiocyanate-bridging between the mercury surface and the Eu³⁺. The diagnostic criteria used to support the proposed mechanism included analysis of the rate—potential behavior and of the effects of competitively adsorbed iodide ions upon the reduction rates.     

Preparation and chain extension of C100dicarboxylic acid

The selective degradation of polyethylene crystals with fuming nitric acid has been used to prepare a C₁₀₀ chain with terminal functional groups. After treating this product with concentrated sulphuric acid, it is shown by average molecular weight, titration and infra-red spectroscopy measurements that each molecule contains two carboxylic acid groups. This C₁₀₀ dicarboxylic acid is chain extended with several low molecular weight difunctional coupling agents. Block copolymers of this acid and poly(propylene glycol) have been prepared.     

Comparison of the structural and chemical composition of giant T-even phage heads.

A study has been made of the structure of the capsids of T4D giant phage produced from mutants in gene 23 and temperature-sensitive mutants in gene 24, and T4D and T2L giant phage formed by the addition of L-canavanine followed by an Larginine chase in the growth medium. All the giant phage capsids have been shown to be built according to the same geometrical architecture. This consists of a near-hexagonal surface net, lattice constant 129.5 A, folded into a left-handed T = 13 prolate icosahedron elongated along one of its fivefold symmetry axes. Their only apparent difference from wild-type T-even phage capsids is their abnormally elongated tubular part. A comparison of the capsomere morphologies and protein compositions of the giant phage capsids showed that all T4D giants are identical but differ from T2L: The T4D capsomere has a complex (6 + 6 + 1)-type morphology, whereas the T2L has a simple 6-type. T2L phage, however, lack two capsid proteins, "soc" and "hoc", present in T4D. The difference in capsomere morphology can therefore be related to the difference in the protein compositions of these two phage. Possible differences between the initiation and means of length regulation of giant phage heads and the aberrant polyheads are discussed.     

Electrochemical behaviour of mono- and oligonucleotides I. Effects of charge and configuration of adenine mononucleotides on their adsorption at the mercury—solution interface

A systematic study of the adsorption and interfacial behaviour of the adenine mono-nucleotides (5′-AMP, 3′-AMP, cyclic 3′,5′-AMP, 5′-ADP and 5′-ATP) and adenosine for comparison at the HMDE has been carried out at pH 3.4 to 3.5. Thus, the N(1) of the adenine moiety is protonated to a major extent.The adsorption was followed by single sweep voltammetry (measurement of the time integral of the reduction peak of the adsorbed adenine moiety) and by a.c. voltammetry (out-of-phase component of the a.c. response being proportional to the differential double layer capacity). In this paper the situation corresponding to a “dilute” adsorption layer existing at low bulk concentrations is studied for various degrees of coverage. The potential dependence of the coverage is of bell shaped type with an extended maximum region around the potential of electrocapillary maximum (E_ecm) of the blank. For the same bulk concentration the coverage decreases in the series AMP, ADP, ATP, i.e. with increasing negative charge of the nucleotide, and at the same time the potential range of adsorption narrows. Among the monophosphates the coverage decreases in the series 3′-AMP, 5′-AMP, cyclic 3′,5′-AMP. The variations are connected with the varying charge of the mononucleotides and with the possibilities for interactions with adjacent molecules in the adsorption layer.At elevated bulk concentrations above a threshold value a substantial increase in coverage occurs around E_ecm as due to strong interactions between the adsorbed base moieties a rather compact film is formed.     

Spectrophotometric determination of germanium with Catechol Violet and cetyltrimethylammonium bromide

A ternary complex between germanium, Catechol Violet (CV) and cetyltrimethylanunoniuni bromide is proposed for the determination of germanium. The stoichiometric ratio Ge:CV is 1:2. Beer's law is obeyed from 0.1 to 1.0 ppm of Ge. The method is highly selective. Interference from Sn(IV), Fe(III), Bi(III), Cr(VI), Mo(VI), V(V) and Sb(III) in mg amounts is eliminated by extracting the germanium into carbon tetrachloride from 9M HC1 and then stripping into water before the photometric determination.     

Tautomeric equilibrium of 1-β-D-ribofuranosyl-2-keto-4-(n-methoxyamino)pyrimidine

The 1-β-D-ribofuranosides of 2-keto-4-(N-methoxyamino)pyrimidine, 2-keto-4-(N-methyl-N-methoxyamino)pyrimidine, and 2-keto-3-methyl-4-(N-methoxyamino)pyrimidine were synthesized, and their pK_a values were determined by spectrophotometry. The pK_a values of the compounds are evidence that the tautomeric equilibrium between the oxime and hydroxyamine forms of 1-β-D-ribofuranosyl-2-keto-4-(N-methoxyamino)pyrimidine in aqueous solutions is shifted to favor the oxime form (K_T?25).