Improving the IR spectra alignment algorithm with spectra deconvolution and combination with Raman or VCD spectroscopy.

The relative stereochemistry of organic molecules can be determined by comparing theoretical and experimental infrared (IR) spectra of all isomers and assessing the best match. For this purpose, we have recently developed the IR spectra alignment (IRSA) algorithm for automated optimal alignment. IRSA provides a set of quantitative metrics to identify the candidate structure that agrees best with the experimental spectrum. While the correct diastereomer could be determined for the tested sets of rigid and flexible molecules, two issues were identified with more complex compounds that triggered further development. First, strongly overlapping peaks in the IR spectrum are not treated adequately in the current IRSA implementation. Second, the alignment of multiple spectra from different sources (e.g. IR and VCD or Raman) can be improved. In this study, we present an in-depth discussion of these points, followed by the description of modifications to the IRSA algorithm to address them. In particular, we introduce the concept of deconvolution of the experimental and theoretical spectra with a set of pseudo-Voigt bands. The pseudo-Voigt bands have a set of parameters, which can be employed in the alignment algorithm, leading to improved scoring functions. We test the modified algorithm on two data sets. The first set contains compounds with IR and Raman spectra measured in this study, and the second set contains compounds with IR and VCD spectra available in the literature. We show that the algorithm is able to determine the correct diastereomer in all cases. The results highlight that vibrational spectroscopy can be a valuable alternative or complementary method to inform about the stereochemistry of compounds, and the performance of the updated IRSA algorithm suggests that it is a powerful tool for quantitative-based spectral assignments in academia and industry.

Graph-convolutional neural networks for (QM)ML/MM molecular dynamics simulations.

To accurately study the chemical reactions in the condensed phase or within enzymes, both quantum-mechanical description and sufficient configurational sampling are required to reach converged estimates. Here, quantum mechanics/molecular mechanics (QM/MM) molecular dynamics (MD) simulations play an important role, providing QM accuracy for the region of interest at a decreased computational cost. However, QM/MM simulations are still too expensive to study large systems on longer time scales. Recently, machine learning (ML) models have been proposed to replace the QM description. The main limitation of these models lies in the accurate description of long-range interactions present in condensed-phase systems. To overcome this issue, a recent workflow has been introduced combining a semi-empirical method (i.e. density functional tight binding (DFTB)) and a high-dimensional neural network potential (HDNNP) in a Δ-learning scheme. This approach has been shown to be capable of correctly incorporating long-range interactions within a cutoff of 1.4 nm. One of the promising alternative approaches to efficiently take long-range effects into account is the development of graph-convolutional neural networks (GCNNs) for the prediction of the potential-energy surface. In this work, we investigate the use of GCNN models - with and without a Δ-learning scheme - for (QM)ML/MM MD simulations. We show that the Δ-learning approach using a GCNN and DFTB as a baseline achieves competitive performance on our benchmarking set of solutes and chemical reactions in water. This method is additionally validated by performing prospective (QM)ML/MM MD simulations of retinoic acid in water and S-adenoslymethionine interacting with cytosine in water. The results indicate that the Δ-learning GCNN model is a valuable alternative for the (QM)ML/MM MD simulations of condensed-phase systems.

Mutanobactin D from the Human Microbiome: Total Synthesis, Configurational Assignment, and Biological Evaluation.

Mutanobactin D is a non-ribosomal, cyclic peptide isolated from Streptococcus mutans and shows activity reducing yeast-to-hyphae transition as well as biofilm formation of the pathogenic yeast Candida albicans. We report the first total synthesis of this natural product, which relies on enantioselective, zinc-mediated 1,3-dipolar cycloaddition and a sequence of cascading reactions, providing the key lipidated γ-amino acid found in mutanobactin D. The synthesis enables configurational assignment, determination of the dominant solution-state structure, and studies to assess the stability of the lipopeptide substructure found in the natural product. The information stored in the fingerprint region of the IR spectra in combination with quantum chemical calculations proved key to distinguishing between epimers of the α-substituted ß-keto amide. Synthetic mutanobactin D drives discovery and analysis of its effect on growth of other members of the human oral consortium. Our results showcase how total synthesis is central for elucidating the complex network of interspecies communications of human colonizers.

Determining the Regiochemistry and Relative Stereochemistry of Small and Druglike Molecules Using an Alignment Algorithm for Infrared Spectra.

The relative stereochemistry and isomeric substitution pattern of organic molecules is typically determined using nuclear magnetic resonance spectroscopy (NMR). However, NMR spectra are sometimes nonconclusive, e.g., if spectra are extremely crowded, coupling patterns are not resolved, or if symmetry reasons preclude interpretation. Infrared spectroscopy (IR) can provide additional information in such cases, because IR represents a molecule comprehensively by depiction of the complete set of its normal vibrations. The challenge is thereby that diastereomers and substitution isomers often give rise to highly similar IR spectra, and visual distinction is insufficient and may be biased. Here we show the adaptation of an alignment algorithm, originally developed for vibrational circular dichroism (VCD) spectroscopy, to automatically match IR spectra and provide a quantitative measure of the goodness of fit, which can be used to distinguish isomers. The performance of the approach is demonstrated for different use cases: diastereomers of flexible druglike molecules, E/Z-isomers, and substitution isomers of pyrazine and naphthalene. It can be applied to IR spectra measured both in the gas phase (gas chromatography IR) and in solution.

Determination of Absolute Stereochemistry of Flexible Molecules Using a Vibrational Circular Dichroism Spectra Alignment Algorithm.

Experimental determination of the absolute stereochemistry of chiral molecules has been a fundamental challenge in natural sciences for decades. Vibrational circular dichroism (VCD) spectroscopy represents an attractive alternative to traditional methods like X-ray crystallography due to the use of molecules in solution. The interpretation of measured VCD spectra and thus the assignment of the absolute configuration relies on quantum-mechanical calculations. While such calculations are straightforward for rigid molecules with a single conformation, the need to estimate the correct conformational ensemble and energy landscape to obtain the appropriate theoretical spectra poses significant challenges for flexible molecules. In this work, we present the development of a VCD spectra alignment (VSA) algorithm to compare theoretical and experimental VCD spectra. The algorithm determines which enantiomer is more likely to reproduce the experimental spectrum and thus allows the correct assignment of the absolute stereochemistry. The VSA algorithm is successfully applied to determine the absolute chirality of highly flexible molecules, including commercial drug substances. Furthermore, we show that the computational cost can be reduced by performing the full frequency calculation only for a reduced set of conformers. The presented approach has the potential to allow the determination of the absolute configuration of chiral molecules in a robust and efficient manner.

Machine Learning with and for Molecular Dynamics Simulations.

From simple clustering techniques to more sophisticated neural networks, the use of machine learning has become a valuable tool in many fields of chemistry in the past decades. Here, we describe two different ways in which we explore the combination of machine learning (ML) and molecular dynamics (MD) simulations. One topic focuses on how the information in MD simulations can be encoded such that it can be used as input to train ML models for the quantitative understanding of molecular systems. The second topic addresses the utilization of machine learning to improve the set-up, interpretation, as well as accuracy of MD simulations.

Regularized by Physics: Graph Neural Network Parametrized Potentials for the Description of Intermolecular Interactions.

Simulations of molecular systems using electronic structure methods are still not feasible for many systems of biological importance. As a result, empirical methods such as force fields (FF) have become an established tool for the simulation of large and complex molecular systems. The parametrization of FF is, however, time-consuming and has traditionally been based on experimental data. Recent years have therefore seen increasing efforts to automatize FF parametrization or to replace FF with machine-learning (ML) based potentials. Here, we propose an alternative strategy to parametrize FF, which makes use of ML and gradient-descent based optimization while retaining a functional form founded in physics. Using a predefined functional form is shown to enable interpretability, robustness, and efficient simulations of large systems over long time scales. To demonstrate the strength of the proposed method, a fixed-charge and a polarizable model are trained on ab initio potential-energy surfaces. Given only information about the constituting elements, the molecular topology, and reference potential energies, the models successfully learn to assign atom types and corresponding FF parameters from scratch. The resulting models and parameters are validated on a wide range of experimentally and computationally derived properties of systems including dimers, pure liquids, and molecular crystals.

Learning Atomic Multipoles: Prediction of the Electrostatic Potential with Equivariant Graph Neural Networks.

The accurate description of electrostatic interactions remains a challenging problem for classical potential-energy functions. The commonly used fixed partial-charge approximation fails to reproduce the electrostatic potential at short range due to its insensitivity to conformational changes and anisotropic effects. At the same time, possibly more accurate machine-learned (ML) potentials struggle with the long-range behavior due to their inherent locality ansatz. Employing a multipole expansion offers in principle an exact treatment of the electrostatic potential such that the long-range and short-range electrostatic interactions can be treated simultaneously with high accuracy. However, such an expansion requires the calculation of the electron density using computationally expensive quantum-mechanical (QM) methods. Here, we introduce an equivariant graph neural network (GNN) to address this issue. The proposed model predicts atomic multipoles up to the quadrupole, circumventing the need for expensive QM computations. By using an equivariant architecture, the model enforces the correct symmetry by design without relying on local reference frames. The GNN reproduces the electrostatic potential of various systems with high fidelity. Possible uses for such an approach include the separate treatment of long-range interactions in ML potentials, the analysis of electrostatic potential surfaces, and static multipoles in polarizable force fields.

Recent developments in multiscale free energy simulations.

Physics-based free energy simulations enable the rigorous calculation of properties, such as conformational equilibria, solvation or binding free energies. While historically most applications have occurred at the atomistic level of resolution, a range of advances in the past years make it possible now to reliably cross the temporal, spatial and theory scales for the modeling of complex systems or the efficient prediction of results at the accuracy level of expensive quantum-mechanical calculations. In this mini-review, we discuss recent methodological advances as well as opportunities opened up by the introduction of machine learning approaches, which tackle the diverse challenges across the different scales, improve the accuracy and feasibility, and push the boundaries of multiscale free energy simulations.

Machine Learning in QM/MM Molecular Dynamics Simulations of Condensed-Phase Systems.

Quantum mechanics/molecular mechanics (QM/MM) molecular dynamics (MD) simulations have been developed to simulate molecular systems, where an explicit description of changes in the electronic structure is necessary. However, QM/MM MD simulations are computationally expensive compared to fully classical simulations as all valence electrons are treated explicitly and a self-consistent field (SCF) procedure is required. Recently, approaches have been proposed to replace the QM description with machine-learned (ML) models. However, condensed-phase systems pose a challenge for these approaches due to long-range interactions. Here, we establish a workflow, which incorporates the MM environment as an element type in a high-dimensional neural network potential (HDNNP). The fitted HDNNP describes the potential-energy surface of the QM particles with an electrostatic embedding scheme. Thus, the MM particles feel a force from the polarized QM particles. To achieve chemical accuracy, we find that even simple systems require models with a strong gradient regularization, a large number of data points, and a substantial number of parameters. To address this issue, we extend our approach to a Δ-learning scheme, where the ML model learns the difference between a reference method (density functional theory (DFT)) and a cheaper semiempirical method (density functional tight binding (DFTB)). We show that such a scheme reaches the accuracy of the DFT reference method while requiring significantly less parameters. Furthermore, the Δ-learning scheme is capable of correctly incorporating long-range interactions within a cutoff of 1.4 nm. It is validated by performing MD simulations of retinoic acid in water and the interaction between S-adenoslymethioniat and cytosine in water. The presented results indicate that Δ-learning is a promising approach for (QM)ML/MM MD simulations of condensed-phase systems.