Fast Pure Shift NMR Spectroscopy Using Attention-Assisted Deep Neural Network.

Pure shift NMR spectroscopy enables the robust probing on molecular structure and dynamics, benefiting from great resolution enhancements. Despite extensive application landscapes in various branches of chemistry, the long experimental times induced by the additional time dimension generally hinder its further developments and practical deployments, especially for multi-dimensional pure shift NMR. Herein, this study proposes and implements the fast, reliable, and robust reconstruction for accelerated pure shift NMR spectroscopy with lightweight attention-assisted deep neural network. This deep learning protocol allows one to regain high-resolution signals and suppress undersampling artifacts, as well as furnish high-fidelity signal intensities along with the accelerated pure shift acquisition, benefitting from the introduction of the attention mechanism to highlight the spectral feature and information of interest. Extensive results of simulated and experimental NMR data demonstrate that this attention-assisted deep learning protocol enables the effective recovery of weak signals that are almost drown in the serious undersampling artifacts, and the distinction and recognition of close chemical shifts even though using merely 5.4% data, highlighting its huge potentials on fast pure shift NMR spectroscopy. As a result, this study affords a promising paradigm for the AI-assisted NMR protocols toward broader applications in chemistry, biology, materials, and life sciences, and among others.

Pure-Shift-Based Proton Magnetic Resonance Spectroscopy for High-Resolution Studies of Biological Samples.

Proton magnetic resonance spectroscopy (1H MRS) presents a powerful tool for revealing molecular-level metabolite information, complementary to the anatomical insight delivered by magnetic resonance imaging (MRI), thus playing a significant role in in vivo/in vitro biological studies. However, its further applications are generally confined by spectral congestion caused by numerous biological metabolites contained within the limited proton frequency range. Herein, we propose a pure-shift-based 1H localized MRS method as a proof of concept for high-resolution studies of biological samples. Benefitting from the spectral simplification from multiplets to singlet peaks, this method addresses the challenge of spectral congestion encountered in conventional MRS experiments and facilitates metabolite analysis from crowded NMR resonances. The performance of the proposed pure-shift 1H MRS method is demonstrated on different kinds of samples, including brain metabolite phantom and in vitro biological samples of intact pig brain tissue and grape tissue, using a 7.0 T animal MRI scanner. This proposed MRS method is readily implemented in common commercial NMR/MRI instruments because of its generally adopted pulse-sequence modules. Therefore, this study takes a meaningful step for MRS studies toward potential applications in metabolite analysis and disease diagnosis.

Accelerated Pure Shift NMR Spectroscopy with Deep Learning.

Pure shift nuclear magnetic resonance (NMR) spectroscopy presents a promising solution to provide sufficient spectral resolution and has been increasingly applied in various branches of chemistry, but the optimal resolution is generally accompanied by long experimental times. We present a proof of concept of deep learning for fast, high-quality, and reliable pure shift NMR reconstruction. The deep learning (DL) protocol allows one to eliminate undersampling artifacts, distinguish peaks with close chemical shifts, and reconstruct high-resolution pure shift NMR spectroscopy along with accelerated acquisition. More meaningfully, the lightweight neural network delivers satisfactory reconstruction performance on personal computers by several hundred simulated data learning, which somewhat lifts the prohibiting demand for a large volume of real training samples and advanced computing hardware generally required in DL projects. Additionally, an M-to-S strategy applicable to common DL cases is further exploited to boost the network generalization capability. As a result, this study takes a meaningful step toward deep learning protocols for broad chemical applications.

Efficient determination of scalar coupling networks by band selective decoupled 2D NMR spectroscopy.

Scalar (J) couplings constitute one of vital features observed in NMR spectroscopy and show valuable information for molecular structure elucidation and conformation analysis. However, existing J coupling measurement techniques are generally confined by the concerns of resolution, SNR, and experimental efficiency. Herein, we exploit an efficient 2D NMR protocol to deal with the above concerns by enabling rapid, sensitive, and high-resolution J coupling extraction. This protocol delivers full-resolved pure shift 2D absorption-mode spectroscopy to gain great convenience for efficient coupling measurements on overcrowded NMR signals. Resulting from band selective signal evolution, this protocol ensures high signal intensity with full magnetization preservation to meet the demand on probing low-concentration samples. This protocol focuses on accessing coupling information between specific two coupled spin families, and it is not applicable to all possible spin systems. Besides, it adopts echo-train selective refocusing acquisition to accelerate pure shift 2D J-edited implementations into pseudo-2D acquisition, and thus holding the experimental efficiency similar to conventional SERF experiments. Therefore, this study presents a promising tool for efficient extraction of J coupling networks, and takes an important step for coupling measurement techniques with wide applications on molecular conformation elucidation and stereochemical configuration analysis.

High-Quality Reconstruction for Laplace NMR Based on Deep Learning.

Laplace nuclear magnetic resonance (NMR) exploits relaxation and diffusion phenomena to reveal information regarding molecular motions and dynamic interactions, offering chemical resolution not accessible by conventional Fourier NMR. Generally, the applicability of Laplace NMR is subject to the performance of signal processing and reconstruction algorithms involving an ill-posed inverse problem. Here, we propose a proof-of-concept of a deep-learning-based method for rapid and high-quality spectra reconstruction from Laplace NMR experimental data. This reconstruction method is performed based on training on synthetic exponentially decaying data, which avoids a vast amount of practically acquired data and makes it readily suitable for one-dimensional relaxation and diffusion measurements by commercial NMR instruments.

Graph Neural Tree: A novel and interpretable deep learning-based framework for accurate molecular property predictions.

Determining various properties of molecules is a critical step in drug discovery. Recently, with the improvement of large heterogeneous datasets and the development of deep learning approaches, more and more scientists have turned their attention to neural network-based virtual preliminary screening to reduce the time and monetary cost of drug discovery. However, the poor interpretability of deep learning masks causality, so models' conclusions are often beyond the comprehension of human users, which reduces the credibility of the model and makes it difficult for chemists to further narrow the huge chemical space based on models' results. Thus, this study develops a novel framework consisting of Graph Neural Networks for feature extraction, Curriculum-Based Learning Strategies for optimization, and a Learning Binary Neural Tree (LBNT) for prediction, to improve the performance of neural networks and reveal their decision-making process to chemists. The framework encodes molecular graph data with graph neural networks (GNNs), then retrains the encoder with curriculum-based learning strategies to reduce uncertainty and improve accuracy, and finally uses LBNT as the predictor, which joint retrains with the encoder after independently training, for prediction and visualization. The framework is validated on the public datasets and compared to single GNNs with normal training strategies as well as GNN encoders with common machine learning predictors instead of the LBNT predictor. The result reveals that the proposed framework enhances the point prediction accuracy of the completely trained GNN and reduces its uncertainty through curriculum-based learning, and further improves the accuracy by combining LBNT. Besides, compared with common machine learning tools, the LBNT predictor generally has the best performance because of joint retraining with the GNN encoder. The decision-making process of LBNT is also better and easier to explain than that of other models.

A Pure Shift-Based Nuclear Magnetic Resonance Method for In-Phase Three-Dimensional Diffusion-Ordered Spectroscopy.

Diffusion-ordered nuclear magnetic resonance spectroscopy (DOSY) plays a vital role in mixture studies. However, its applications to complex mixture samples are generally limited by spectral congestion along the chemical shift domain caused by extensive J coupling networks and abundant compounds. Herein, we develop the in-phase multidimensional DOSY strategy for complex mixture analyses by simultaneously revealing molecular self-diffusion behaviors and multiplet structures with optimal spectral resolution. As a proof of concept, two pure shift-based three-dimensional (3D) DOSY protocols are proposed to record high-resolution 3D spectroscopic view with separated mixture components and their resolved multiplet coupling structures, thus suitable for analyzing complex mixtures that contain abundant compounds and complicated molecular structures, even under adverse magnetic field conditions. Therefore, this study shows a promising tool for component analyses and multiplet structure studies on practical mixture samples.

Adaptable Singlet-Filtered Nuclear Magnetic Resonance Spectroscopy for Chemical and Biological Applications.

Proton nuclear magnetic resonance (1H NMR) spectroscopy presents a powerful detection tool for studying chemical compositions and molecular structures. In practical chemical and biological applications, 1H NMR experiments are generally confronted with the challenge of spectral congestions caused by abundant observable components and intrinsic limitations of a narrow frequency distribution range and extensive J coupling splitting. Herein, a one-dimensional (1D) general NMR method is proposed to individually extract the signals of targeted proton groups based on their endogenous spin singlet states excited from J coupling interactions, and it is suitable for high-resolution detections on complex chemical and biological samples. The applicability of the proposed method is demonstrated by experimental observations on chemical solutions containing different coupled components, intact grape tissues subjected to crowded resonances, and in vitro pig brain with various metabolites. Moreover, the proposed method is further exploited for magnetic resonance spectroscopy applications by directly combining the spatial localization module, showing promise in in vivo biological metabolite studies.

Simultaneous determination of multiple coupling networks by high-resolution 2D J-edited NMR spectroscopy.

J coupling constitutes an important NMR parameter for molecular-level composition analysis and conformation elucidation. Dozens of J-based approaches have been exploited for J coupling measurement and coupling network determination, however, they are generally imposed to insufficient spectral resolution to resolve crowded NMR resonances and low measurement efficiency that a single experiment records one J coupling network. Herein, we propose a general NMR method to collect high-resolution 2D J-edited NMR spectra, which are characterized with advantages of pure absorptive lineshapes, decoupled chemical shift dimension, as well as eliminated axial peaks, thus facilitating J coupling partner assignments and J coupling constant measurements. More meaningfully, this protocol allows simultaneous determination of multiple coupling networks for highly efficient multiplet analyses via addressing multiple protons within one single experiment. Additionally, another variant is proposed for high-resolution applications under adverse magnetic field conditions. Therefore, this study provides a useful NMR protocol for configurational and structural studies with extensive applications in chemistry, biology, and material science.

Multiplet analysis by strong-coupling-artifact-suppression 2D J-resolved NMR spectroscopy.

Benefitting from the capability of recording scalar (J) couplings and bonding information, 2D J-resolved NMR spectroscopy constitutes an important tool for molecular structure analysis and mixture component identification. Unfortunately, conventional 2D J-resolved experiments generally encounter challenges of insufficient spectral resolution and strong coupling artifacts. In this study, a general NMR approach is exploited to record absorption-mode artifact-free 2D J-resolved spectra. This proposal adopts the advanced triple-spin-echo pure shift yielded by chirp excitation element to eliminate J coupling splittings and preserve chemical shifts along the F2 dimension, and it additionally utilizes the echo-train J acquisition to reveal the multiplet structure along the F1 dimension in accelerated experimental acquisition. Thus, it permits one to extract multiplet structure information from crowded spectral regions in one-shot experiments, with considerable resolution advantage resulting from completely decoupling F2 dimension and absorption-mode presentation, thus facilitating analysis on complex samples. More importantly, this method grants the superior performance on suppressing strong coupling artifacts, which have been affirmed by experiments on a series of chemical samples. As a consequence, this proposed method serves as a useful tool for J coupling measurements and multiplet structure analyses on complex samples that contain crowded NMR resonances and strong coupling spin systems, and it may exhibit broad application potentials in fields of physics, chemistry, and medical science, among others.

High-Resolution Reconstruction for Multidimensional Laplace NMR.

As a perfect complement to conventional NMR that aims for chemical structure elucidation, Laplace NMR constitutes a powerful technique to study spin relaxation and diffusion, revealing information on molecular motions and spin interactions. Different from conventional NMR adopting Fourier transform to deal with the acquired data, Laplace NMR relies on specially designed signal processing and reconstruction algorithms resembling the inverse Laplace transform, and it generally faces severe challenges in cases where high spectral resolution and high spectral dimensionality are required. Herein, based on the tensor technique for high-dimensional problems and the sparsity assumption, we propose a general method for high-resolution reconstruction of multidimensional Laplace NMR data. We show that the proposed method can reconstruct multidimensional Laplace NMR spectra in a high-resolution manner for exponentially decaying relaxation and diffusion data acquired by commercial NMR instruments. Therefore, it would broaden the scope of multidimensional Laplace NMR applications.

Diffusion Analysis on Complex Mixtures under Adverse Magnetic Field Conditions by Spatially-Selective Pure Shift-Based DOSY.

Diffusion-ordered NMR spectroscopy (DOSY) serves as a noninvasive spectroscopic method for studying intact mixtures and identifying individual components present in mixtures according to their diffusion behaviors. However, DOSY techniques generally fail to discriminate complex compositions which exhibit crowded or overlapped NMR signals, particularly under adverse magnetic field conditions. Herein, we exploit the spatially selective pure shift-based DOSY strategy to address this challenge by eliminating inhomogeneous line broadenings and extracting pure shift singlets, thereby expediting diffusion analyses on complex mixtures. More importantly, this strategy is further applied to observing and analyzing electro-oxidation processes of blended alcohols, suggesting its potential to monitoring in situ electrochemical reactions. This study demonstrates a meaningful NMR trial for diffusion analysis on complex mixtures under adverse experimental circumstances, and particularly, it provides a proof-of-concept technique for electrochemical studies and shows promising prospects for applications in chemistry, biology, energy, etc.

Highly Efficient Determination of Complex NMR Multiplet Structures in Inhomogeneous Magnetic Fields.

Proton-proton scalar (J) coupling plays an important role in disentangling molecular structures and spatial conformations. But it is challenging to extract J coupling networks from congested 1H NMR spectra, especially in inhomogeneous magnetic fields. Herein, we propose a general liquid NMR protocol, named HR-G-SERF, to implement highly efficient determination of individual J couplings and corresponding coupling networks via simultaneously suppressing effects of spectral congestions and magnetic field inhomogeneity. This method records full-resolved 2D absorption-mode spectra to deliver great convenience for multipet analyses on complex samples. More meaningfully, it is capable of disentangling multiplet structures of biological samples, that is, grape sarcocarp, despite of its heterogeneous semisolid state and extensive compositions. In addition, a modification, named AH-G-SERF, is developed to compress experimental acquisition and subsequently improve unit-time SNR, while maintaining satisfactory spectral performance. This accelerated variant may further boost the applicability for rapid NMR detections and afford the possibility of adopting hyperpolarized substances to enhance the overall sensitivity. Therefore, this study provides a promising tool for molecular structure elucidations and composition analyses in chemistry, biochemistry, and metabonomics among others.

NMR Relaxation Measurements on Complex Samples Based on Real-Time Pure Shift Techniques.

Longitudinal spin-lattice relaxation (T1) and transverse spin-spin relaxation (T2) reveal valuable information for studying molecular dynamics in NMR applications. Accurate relaxation measurements from conventional 1D proton spectra are generally subject to challenges of spectral congestion caused by J coupling splittings and spectral line broadenings due to magnetic field inhomogeneity. Here, we present an NMR relaxation method based on real-time pure shift techniques to overcome these two challenges and achieve accurate measurements of T1 and T2 relaxation times from complex samples that contain crowded NMR resonances even under inhomogeneous magnetic fields. Both theoretical analyses and detailed experiments are performed to demonstrate the effectiveness and ability of the proposed method for accurate relaxation measurements on complex samples and its practicability to non-ideal magnetic field conditions.

High-Resolution Probing of Heterogeneous Samples by Spatially Selective Pure Shift NMR Spectroscopy.

Liquid NMR spectroscopy generally encounters two major challenges for high-resolution measurements of heterogeneous samples, namely, magnetic field inhomogeneity caused by spatial variations in magnetic susceptibility and spectral congestion induced by crowded NMR resonances. In this study, we demonstrate a spatially selective pure shift NMR approach for high-resolution probing of heterogeneous samples by suppressing effects of field inhomogeneity and J coupling simultaneously. A Fourier phase encoding strategy is proposed and implemented for spatially selective pure shift experiments to enhance signal intensity and further boost the applicability. The spatially selective pure shift method can serve as an effective tool for high-resolution probing of heterogeneous samples, thus presenting interesting prospects for extensive applications in the fields of chemistry, physics, biology, and food science.

A pure shift and spin echo based approach for high-resolution diffusion-ordered NMR spectroscopy.

Diffusion-ordered NMR spectroscopy (DOSY) can be used for separating mixture components according to their individual diffusion behaviors, thus offering a powerful tool for the analysis of compound mixtures. However, conventional DOSY experiments generally encounter the problem of limited resolution in the spectral domain, particularly for applications to complex mixtures that contains crowed resonances in 1D NMR. In addition, chemical exchange effects, bringing about spurious component signals, pose another limitation for interpreting DOSY measurements. Here, a general DOSY method is proposed based on pure shift extraction and spin echo evolution to obtain high-resolution 2D DOSY spectra, along with the suppression on effects of chemical exchange and J coupling. Both theoretical analyses and experimental results suggest that the proposed method is useful for high-resolution DOSY measurements on complex mixtures that contains crowded or even overlapped NMR resonances and exchanging spin systems.

General Two-Dimensional Absorption-Mode J-Resolved NMR Spectroscopy.

Two-dimensional (2D) J-resolved NMR technique offers a natural solution for disentangling complex mixtures that suffer from crowded spectra in 1D NMR. The applicability of classical 2D J-resolved spectroscopy is inevitably limited by phase-twist lineshapes and strong coupling artifacts. Here, a general and robust NMR method is proposed to record 2D absorption-mode J-resolved spectra in rapid acquisition manner. This method can also reduce the impact of strong coupling artifacts, thus achieving full considerations for applications. Intuitively, this method delivers pure chemical shifts along one dimension and orthogonally adds J couplings along the other dimension, free of 45° spectral shearing. It may provide a powerful tool for structural and configurational studies as well as biological analyses.

Ultrahigh-Resolution NMR Spectroscopy for Rapid Chemical and Biological Applications in Inhomogeneous Magnetic Fields.

NMR spectroscopy is a commonly used analytical technique in practical applications, and its applicability is further promoted by pure chemical shift techniques based on spectral simplification for analyses. Unfortunately, magnetic field inhomogeneity caused by adverse experimental conditions remains an obstacle restricting NMR applications. In this study, we introduce a new NMR method for high-resolution pure shift proton (1H) NMR measurements in inhomogeneous magnetic fields. We demonstrate that the method allows one to perform chemical analyses on complex solutions in deshimmed magnetic fields, to obtain metabolite information on intact biological tissues with intrinsic field inhomogeneities and to achieve in situ electrochemical detection under externally adverse field conditions. This approach is readily implemented on common commercial NMR instruments without field shimming and locking procedures, specialized hardware requirements as well as complicated sample pretreatments. It provides an effective tool for NMR applications to high-resolution chemical and biological measurements under inhomogeneous magnetic field conditions.