Spectroscopy from Machine Learning by Accurately Representing the Atomic Polar Tensor

Vibrational spectroscopy is a key technique to elucidate microscopic structure and dynamics. Without the aid of theoretical approaches, it is, however, often difficult to understand such spectra at a microscopic level. Ab initio molecular dynamics has repeatedly proved to be suitable for this purpose; however, the computational cost can be daunting. Here, the E(3)-equivariant neural network e3nn is used to fit the atomic polar tensor of liquid water a posteriori on top of existing molecular dynamics simulations. Notably, the introduced methodology is general and thus transferable to any other system as well. The target property is most fundamental and gives access to the IR spectrum, and more importantly, it is a highly powerful tool to directly assign IR spectral features to nuclear motion─a connection which has been pursued in the past but only using severe approximations due to the prohibitive computational cost. The herein introduced methodology overcomes this bottleneck. To benchmark the machine learning model, the IR spectrum of liquid water is calculated, indeed showing excellent agreement with the explicit reference calculation. In conclusion, the presented methodology gives a new route to calculate accurate IR spectra from molecular dynamics simulations and will facilitate the understanding of such spectra on a microscopic level.

Vibrational Circular Dichroism from DFT Molecular Dynamics: The AWV Method

The paper illustrates the Activity Weighted Velocities (AWV) methodology to compute Vibrational Circular Dichroism (VCD) anharmonic spectra from Density Functional Theory (DFT) molecular dynamics. AWV calculates the spectra by the Fourier Transform of the time correlation functions of velocities, weighted by specific observables: the Atomic Polar Tensors (APTs) and the Atomic Axial Tensors (AATs). Indeed, AWV shows to correctly reproduce the experimental spectra for systems in the gas and liquid phases, both in the case of weakly and strongly interacting systems. The comparison with the experimental spectra is striking especially in the fingerprint region, as demonstrated by the three benchmark systems discussed: (1S)-Fenchone in the gas phase, (S)-(−)-Propylene oxide in the liquid phase, and (R)-(−)-2-butanol in the liquid phase. The time evolution of APTs and AATs can be adequately described by a linear combination of the tensors of a small set of appropriate reference structures, strongly reducing the computational cost without compromising accuracy. Additionally, AWV allows the partition of the spectral signal in its molecular components without any expensive postprocessing and any localization of the charge density or the wave function.

Some opinions on MD-based vibrational spectroscopy of gas phase molecules and their assembly: An overview of what has been achieved and where to go

We hereby review molecular dynamics simulations for anharmonic gas phase spectroscopy and provide some of our opinions of where the field is heading. With these new directions, the theoretical IR/Raman spectroscopy of large (bio)-molecular systems will be more easily achievable over longer time-scale MD trajectories for an increase in accuracy of the MD-IR and MD-Raman calculated spectra. With the new directions presented here, the high throughput 'decoding' of experimental IR/Raman spectra into 3D-structures should thus be possible, hence advancing e.g. the field of MS-IR for structural characterization by spectroscopy. We also review the assignment of vibrational spectra in terms of anharmonic molecular modes from the MD trajectories, and especially introduce our recent developments based on Graph Theory algorithms. Graph Theory algorithmic is also introduced in this review for the identification of the molecular 3D-structures sampled over MD trajectories.

Probing the formation of isolated cyclo-FF peptide clusters by far-infrared action spectroscopy

Small cyclic peptides containing phenylalanine residues are prone to aggregate in the gas phase into highly hydrophobic chains. A combination of laser desorption, mass spectrometry and conformational selective IR-UV action spectroscopy allows us to obtain detailed structural insights into the formation processes of the cyclic L-phenylalanyl-L-phenylalanine dipeptide (named cyclo-FF) aggregates. The rigid properties of cyclo-FF result in highly resolved IR spectra for the smaller clusters (n ≤ 3) and corresponding conformational assignments. For the higher order clusters (n > 3) the spectra are less resolved, however the observed ratios, peak positions and trends in IR shifts are key to make predictions on their structural details. Whereas the mid-IR spectral region between 1000-1800 cm^-1 turns out to be undiagnostic for these small aggregates and the 3 μm region only for specific calculated structures, the far-IR contains valuable information that allows for clear assignments.

Determination of kinetic properties in unimolecular dissociation of complex systems from graph theory based analysis of an ensemble of reactive trajectories

In this paper, we report how graph theory can be used to analyze an ensemble of independent molecular trajectories, which can react during the simulation time-length, and obtain structural and kinetic information. This method is totally general and here is applied to the prototypical case of gas phase fragmentation of protonated cyclo-di-glycine. This methodology allows us to analyze the whole set of trajectories in an automatic computer-based way without the need of visual inspection but by getting all the needed information. In particular, we not only determine the appearance of different products and intermediates but also characterize the corresponding kinetics. The use of colored graph and canonical labeling allows for the correct characterization of the chemical species involved. In the present case, the simulations consist of an ensemble of unimolecular fragmentation trajectories at constant energy such that from the rate constants at different energies, the threshold energy can also be obtained for both global and specific pathways. This approach allows for the characterization of ion-molecule complexes, likely through a roaming mechanism, by properly taking into account the elusive nature of such species. Finally, it is possible to directly obtain the theoretical mass spectrum of the fragmenting species if the reacting system is an ion as in the specific example.

Ultrafast Transient Infrared Spectroscopy of Photoreceptors with Polarizable QM/MM Dynamics

Ultrafast transient infrared (TRIR) spectroscopy is widely used to measure the excitation-induced structural changes of protein-bound chromophores. Here, we design a novel and general strategy to compute TRIR spectra of photoreceptors by combining μs-long MM molecular dynamics with ps-long QM/AMOEBA Born-Oppenheimer molecular dynamics (BOMD) trajectories for both ground and excited electronic states. As a proof of concept, the strategy is here applied to AppA, a blue-light-utilizing flavin (BLUF) protein, found in bacteria. We first analyzed the short-time evolution of the embedded flavin upon excitation revealing that its dynamic Stokes shift is ultrafast and mainly driven by the internal reorganization of the chromophore. A different normal-mode representation was needed to describe ground- and excited-state IR spectra. In this way, we could assign all of the bands observed in the measured transient spectrum. In particular, we could characterize the flavin isoalloxazine-ring region of the spectrum, for which a full and clear description was missing.

Functional Group Identification for FTIR Spectra Using Image-Based Machine Learning Models

Fourier transform infrared spectroscopy (FTIR) is a ubiquitous spectroscopic technique. Spectral interpretation is a time-consuming process, but it yields important information about functional groups present in compounds and in complex substances. We develop a generalizable model via a machine learning (ML) algorithm using convolutional neural networks (CNNs) to identify the presence of functional groups in gas-phase FTIR spectra. The ML models reduce the amount of time required to analyze functional groups and facilitate interpretation of FTIR spectra. Through web scraping, we acquire intensity-frequency data from 8728 gas-phase organic molecules within the NIST spectral database and transform the data into spectral images. We successfully train models for 15 of the most common organic functional groups, which we then determine via identification from previously untrained spectra. These models serve to expand the application of FTIR measurements for facile analysis of organic samples. Our approach was done such that we have broad functional group models that infer in tandem to provide full interpretation of a spectrum. We present the first implementation of ML using image-based CNNs for predicting functional groups from a spectroscopic method.

Neutral Peptides in the Gas Phase: Conformation and Aggregation Issues

Combined IR and UV laser spectroscopic techniques in molecular beams merged with theoretical approaches have proven to be an ideal tool to elucidate intrinsic structural properties on a molecular level. It offers the possibility to analyze structural changes, in a controlled molecular environment, when successively adding aggregation partners. By this, it further makes these techniques a valuable starting point for a bottom-up approach in understanding the forces shaping larger molecular systems. This bottom-up approach was successfully applied to neutral amino acids starting around the 1990s. Ever since, experimental and theoretical methods developed further, and investigations could be extended to larger peptide systems. Against this background, the review gives an introduction to secondary structures and experimental methods as well as a summary on theoretical approaches. Vibrational frequencies being characteristic probes of molecular structure and interactions are especially addressed. Archetypal biologically relevant secondary structures investigated by molecular beam spectroscopy are described, and the influences of specific peptide residues on conformational preferences as well as the competition between secondary structures are discussed. Important influences like microsolvation or aggregation behavior are presented. Beyond the linear α-peptides, the main results of structural analysis on cyclic systems as well as on β- and γ-peptides are summarized. Overall, this contribution addresses current aspects of molecular beam spectroscopy on peptides and related species and provides molecular level insights into manifold issues of chemical and biochemical relevance.

Gas-Phase Infrared Spectroscopy of Neutral Peptides: Insights from the Far-IR and THz Domain

Gas-phase, double resonance IR spectroscopy has proven to be an excellent approach to obtain structural information on peptides ranging from single amino acids to large peptides and peptide clusters. In this review, we discuss the state-of-the-art of infrared action spectroscopy of peptides in the far-IR and THz regime. An introduction to the field of far-IR spectroscopy is given, thereby highlighting the opportunities that are provided for gas-phase research on neutral peptides. Current experimental methods, including spectroscopic schemes, have been reviewed. Structural information from the experimental far-IR spectra can be obtained with the help of suitable theoretical approaches such as dynamical DFT techniques and the recently developed Graph Theory. The aim of this review is to underline how the synergy between far-IR spectroscopy and theory can provide an unprecedented picture of the structure of neutral biomolecules in the gas phase. The far-IR signatures of the discussed studies are summarized in a far-IR map, in order to gain insight into the origin of the far-IR localized and delocalized motions present in peptides and where they can be found in the electromagnetic spectrum.

Evaluation of Molecular Polarizability and of Intensity Carrying Modes Contributions in Circular Dichroism Spectroscopies

We re-examine the theory of electronic and vibrational circular dichroism spectroscopy in terms of the formalism of frequency-dependent molecular polarizabilities. We show the link between Fermi’s gold rule in circular dichroism and the trace of the complex electric dipole–magnetic dipole polarizability. We introduce the C++ code polar to compute the molecular polarizability complex tensors from quantum chemistry outputs, thus simulating straightforwardly UV-visible absorption (UV-Vis)/electronic circular dichroism (ECD) spectra, and infrared (IR)/vibrational circular dichroism (VCD) spectra. We validate the theory and the code by referring to literature data of a large group of chiral molecules, showing the remarkable accuracy of density functional theory (DFT) methods. We anticipate the application of this methodology to the interpretation of vibrational spectra in various measurement conditions, even in presence of metal surfaces with plasmonic properties. Our theoretical developments aim, in the long run, at embedding the quantum-mechanical details of the chiroptical spectroscopic response of a molecule into the simulation of the electromagnetic field distribution at the surface of plasmonic devices. Such simulations are also instrumental to the interpretation of the experimental spectra measured from devices designed to enhance chiroptical interactions by the surface plasmon resonance of metal nanostructures.

Formation of Neutral Peptide Aggregates as Studied by Mass-Selective IR Action Spectroscopy

The spontaneous aggregation of proteins and peptides is widely studied owing to its relation to neurodegenerative diseases. To understand the underlying principles of peptide aggregation, elucidation of structure and structural changes upon their formation is key. This level of detail can be obtained by studying the peptide self‐assembly in the gas phase. Structural characterization of aggregates is mainly done on charged species, as adding charges is an intrinsic part of the technique to bring molecules into the gas phase. Studying neutral peptide aggregates will complement the existing picture. These studies are restricted to dimers due to experimental limitations. Herein, we present advances in laser desorption molecular beam spectroscopy to form neutral peptide aggregates consisting of up to 14 monomeric peptides in the gas phase. The combination of this technique with IR–UV spectroscopy allowed us to select each aggregate by size and subsequently characterize its structure.