NMR spectroscopy is ubiquitous in structural elucidation of synthetic compounds, metabolites, natural products and materials in chemistry, as well as structural and functional characterisation of biomolecules and macromolecular complexes. The versatility of NMR spectroscopy derives from multiple-pulse experiments, where nuclear correlations are encoded in multidimensional spectra. However, the direct result of an NMR experiment is not a spectrum, but a time series. The NMR spectrum is generated from the time response of the pulsed experiment through the application of a method for spectrum analysis, which constructs a frequency-domain spectrum from, or consistent with, the time-domain empirical data. Signal processing and pulsed NMR therefore go hand-in-hand in modern NMR spectroscopy. The inherently weak NMR signal has made signal processing a vital step in the varied applications of NMR. Basic understanding of signal processing is therefore a pre-requisite for the modern NMR spectroscopist.
In recent years we have witnessed an explosion in the variety of methods for spectrum analysis employed in NMR, motivated by limitations of the discrete Fourier transform (DFT) that was seminal in the development of modern pulsed NMR experiments. Prime among these limitations is the difficulty (using the DFT) of obtaining high-resolution spectra from short data records. An inherent limitation of the DFT is the requirement that data be collected at uniform time intervals; many modern methods of spectrum analysis circumvent this requirement to enable much more efficient sampling approaches. Other modern methods of spectrum analysis obtain high-resolution spectra by implicitly or explicitly modelling the NMR signals. Alternatively, we have witnessed the development of approaches that collect multidimensional data via multiplexing in space—exploiting the physical dimensions of the sample—rather than via sampling a series of indirect time dimensions, or approaches that tailor the pulse sequence in ways that enable drastically faster sampling in time. Together, methods based on nonuniform sampling in time or sampling in space enable a class of experiments described as Fast NMR Data Acquisition. The methods that increase the speed of data acquisition through non-conventional pulse sequence design include SOFAST-NMR and Single Scan NMR, covered in the first two chapters of this book. Methods based on modelling the signal to obtain high resolution spectra from short data records or those that support nonuniform sampling are described in Chapters 3–4 and Chapters 5–10, respectively. The latter are further categorized by those that sample in a deterministic matter, i.e. uniformly along radial or concentric patterns (Chapters 5 and 6) or those that seek incoherence in the distribution of the sampling times (Chapters 7–10).
In this book we have brought together contributions from leading scientists in the development of Fast NMR Data Acquisition to provide a comprehensive reference text on this rapidly growing field. The popularity and rapid expansion of fast acquisition methods is evident in the literature. For example, a search for non-uniform sampling (NUS) terms (non-uniform, non-linear, projection, radial, etc.) and NMR revealed 185 publications since 2000 (Scopus). 13 of these were published between 2000 and 2005, when projection reconstruction and reduced dimensionality experiments were being developed. In 2005–2010 the impact of these experiments and their relationship to data sampling led to 44 publications, and in 2010–2015 the field further expanded with 105 publications, with the introduction of various “compressed sensing” techniques and the elucidation of their relationship to established methods. Similarly, citations of these articles have risen from 237 citations in 2010 to ∼800 citations in 2015. These numbers, although crude, nevertheless show how interest in fast acquisition techniques has exploded over the past two decades, moving from relative obscurity to the mainstream. The widespread adoption of Fast Acquisition Methods is perhaps most evident in the rapid adaptation of modern spectrometers to these methods. In 2005, purpose-written pulse sequences had to be used to perform NUS, whilst today it is treated as simply another standard acquisition parameter during experimental setup by most commercial NMR instruments.
There is no doubt that fast acquisition methods are now firmly established as a part of modern NMR spectroscopy and we hope that this text book will serve to orient the spectroscopist in this new era, providing improved understanding of the many methods on offer and enabling informed decisions on how to make the most of the faint nuclear signals to resolve complex chemical and biological problems.
We are deeply indebted to our colleagues who contributed to this volume, and we express special thanks to Professor Lucio Frydman for contributing his perspective in the Foreword. We are also grateful to the editorial staff of the Royal Society of Chemistry for their enthusiasm for this project and their tireless efforts during editing and production. Finally, MM wishes to acknowledge support from the Australian Research Council in establishing fast acquisition methods towards the automation of protein structure determination by NMR (FTl10100925). JCH wishes to acknowledge the generous support of the US National Institutes of Health via the grant P41GM111135, which enabled the establishment of NMRbox.org: National Center for Biomolecular NMR Data Processing and Analysis. All of the authors of computer codes for non-Fourier methods represented in this book have generously consented to distributing their software via NMRbox.org.
Mehdi Mobli and Jeffrey C. Hoch