MolE8: finding DFT potential energy surface minima values from force-field optimised organic molecules with new machine learning representations

The use of machine learning techniques in computational chemistry has gained significant momentum since large molecular databases are now readily available. Predictions of molecular properties using machine learning have advantages over the traditional quantum mechanics calculations because they can be cheaper computationally without losing the accuracy. We present a new extrapolatable and explainable molecular representation based on bonds, angles and dihedrals that can be used to train machine learning models. The trained models can accurately predict the electronic energy and the free energy of small organic molecules with atom types C, H N and O, with a mean absolute error of 1.2 kcal mol⁻¹. The models can be extrapolated to larger organic molecules with an average error of less than 3.7 kcal mol⁻¹ for 10 or fewer heavy atoms, which represent a chemical space two orders of magnitude larger. The rapid energy predictions of multiple molecules, up to 7 times faster than previous ML models of similar accuracy, has been achieved by sampling geometries around the potential energy surface minima. Therefore, the input geometries do not have to be located precisely on the minima and we show that accurate density functional theory energy predictions can be made from force-field optimised geometries with a mean absolute error 2.5 kcal mol⁻¹.

New representations and machine learning calculate DFT minima from force field geometries.     

DP4-AI automated NMR data analysis: straight from spectrometer to structure

A robust system for automatic processing and assignment of raw ¹³C and ¹H NMR data DP4-AI has been developed and integrated into our computational organic molecule structure elucidation workflow. Starting from a molecular structure with undefined stereochemistry or other structural uncertainty, this system allows for completely automated structure elucidation. Methods for NMR peak picking using objective model selection and algorithms for matching the calculated ¹³C and ¹H NMR shifts to peaks in noisy experimental NMR data were developed. DP4-AI achieved a 60-fold increase in processing speed, and near-elimination of the need for scientist time, when rigorously evaluated using a challenging test set of molecules. DP4-AI represents a leap forward in NMR structure elucidation and a step-change in the functionality of DP4. It enables high-throughput analyses of databases and large sets of molecules, which were previously impossible, and paves the way for the discovery of new structural information through machine learning. This new functionality has been coupled with an intuitive GUI and is available as open-source software at https://github.com/KristapsE/DP4-AI.

A robust system for automatic processing and assignment of raw ¹³C and ¹H NMR data DP4-AI has been developed and integrated into our computational organic molecule structure elucidation workflow.     

NMR Spectroscopic Assignment of Backbone and Side‐Chain Protons in Fully Protonated Proteins: Microcrystals,Sedimented Assemblies,and Amyloid Fibrils

We demonstrate sensitive detection of alpha protons of fully protonated proteins by solid‐state NMR spectroscopy with 100–111 kHz magic‐angle spinning (MAS). The excellent resolution in the Cα‐Hα plane is demonstrated for 5 proteins, including microcrystals, a sedimented complex, a capsid and amyloid fibrils. A set of 3D spectra based on a Cα–Hα detection block was developed and applied for the sequence‐specific backbone and aliphatic side‐chain resonance assignment using only 500 μg of sample. These developments accelerate structural studies of biomolecular assemblies available in submilligram quantities without the need of protein deuteration.     

Imidazolium and pyridinium salts – solvents influencing the rate and direction of the Fries,Beckmann, and Claisen rearrangements

The Fries, Beckmann, and Claisen rearrangements have been investigated in ionic liquid media. The effect of the structural elements of the latter on the direction of these rearrangements and the product yields has been studied.     

Interlaboratory comparison for quantitative primary metabolite profiling in Pichia pastoris

For the first time, an interlaboratory comparison was performed in the field of quantitative metabolite profiling in Pichia pastoris. The study was designed for the evaluation of different measurement platforms integrating different quantification strategies using internal standardization. Nineteen primary metabolites including amino acids and organic acids were selected for the study. Homogenous samples were obtained from chemostat fermentations after rapid sampling, quenching and filtration, and hot ethanol extraction. Laboratory 1 (BOKU) employed an in vivo-synthesized fully labeled U¹³C cell extracts of P. pastoris for immediate internal standardization upon cell extraction. Quantification was carried out using orthogonal reversed-phase (RP-LC) and hydrophilic interaction chromatography (HILIC) in combination with tandem mass spectrometry. Laboratory 2 (Biocrates) applied a metabolomics kit allowing fully automated, rapid derivatization, solid phase extraction and internal standardization in 96-well plates with immobilized isotopically enriched internal standards in combination with HILIC-MS-MS and RP-LC-MS-MS for organic acids and derivatized amino acids, respectively. In this study, the obtained intracellular concentrations ranged from 0.2 to 108 μmol?g^?1 cell dry weight. The total combined uncertainty was estimated including uncertainty contributions from the corresponding MS-based measurement and sample preparation for each metabolite. Evidently, the uncertainty contribution of sample preparation was lower for the values obtained by laboratory 1, implementing isotope dilution upon extraction. Total combined uncertainties (K?=?2) ranging from 21 to 48 % and from 30 to 57 % were assessed for the quantitative results obtained in laboratories 1 and 2, respectively. The major contribution arose from sample preparation, hence from repeatability precision of the extraction procedure. Finally, the laboratory intercomparison was successful as most of the investigated metabolites showed concentration levels agreeing within their total combined uncertainty, implying that accurate quantification was given. The application of isotope dilution upon extraction was an absolute prerequisite for the quantification of the redox-sensitive amino acid methionine, where no agreement between the two laboratories could be achieved.     

Alkylation of the ambident indole ion in ionic liquids

Alkylation of the ambident indole anion in ionic liquids has been investigated. The reaction rate is greater in ionic liquids than in organic solvents. The polarity of certain ionic liquids has been determined to be located between methanol and acetonitrile. __________ Translated from Khimiya Geterotsiklicheskikh Soedinenii, No. 5, pp. 676–690. May, 2008.     

Benzo[b]thiophen-3(2H)-one 1,1-dioxide—a versatile reagent in the synthesis of spiroheterocycles

New applications of benzo[b]thiophen-3(2H)-one 1,1-dioxide have been investigated. The synthesis of a new heterocyclic system 3H,2′H-spiro[benzo[b]thieno[3,2-b]pyridine-3,2′-benzo[b]thiophene] is described and a mechanism for the cyclisation is proposed.     

Alkylation of ambident indole anion in ionic liquids

Alkylation of indole salts in different ionic liquids is reported. Ionic liquids increase the alkylation reaction rate of ambident indole anion and reduce the effects of counter ions and/or additives, the alkylation reaction rates being independent of the presence of small amounts of protic solvents or water.