NMR Data Processing

Advances in superconducting magnets, pulse sequences, and cryogenic probes have fueled the application of NMR spectroscopy to biomacromolecules. NMR data processing, has advanced little since the early 1980's. The phenomenal increase in the power (and decrease in the cost) of computers since then has created new opportunities to bring high-performance computing to bear to enhance sensitivity, increase resolution, and decrease measuring time.

Well-known limitations of the discrete Fourier transform (DFT) are commonly ameliorated using linear-prediction (LP) extrapolation. We showed that LP extrapolation has subtle but significant defects, introducing frequency errors and false peaks. Maximum-entropy (MaxEnt) reconstruction is a more robust alternative. Because it does not require data to be sampled at uniformly-spaced time intervals, MaxEnt reconstruction permits reduced data collection, or improved sensitivity and resolution for a given measuring time. It is also a powerful method for deconvolution, enabling vitual decoupling. The Rowland NMR Toolkit, our NMR data processing software package, is used by more than 80 laboratories world-wide, and is the basis for the book "NMR Data Processing (Wiley-Liss, 1996).

Documentation and information on how to acquire The Rowland NMR Toolkit can be found here.

Aims: Despite its advantages, a number of challenges have prevented wide-scale adoption of MaxEnt reconstruction in NMR. Foremost are the adjustable parameters, which are not intuitive. We aim to derive a theoretical basis for determining the values of the parameters and to develop algorithms to automate MaxEnt reconstruction. In addition we plan to develop strategies that utilize the nonuniform sampling capabilities of MaxEnt to improve sensitivity and resolution of multidimensional experiments, and to exploit its ability to handle non-Lorentzian lineshapes in solid-state NMR.