One of the most important applications for 2- and higher dimensional NMR spectroscopy is the determination of accurate 3-dimensional structures of small proteins, nucleic acids, and other important biological molecules in solution. this new application, based in large part on NOESY data, offers unique promise for structural understanding of complex biomolecular systems. There are significant challenges however, particularly in the area of computational methods which encompass: (1) optimal preparation of the multi-dimensional NMR data for utilization including quantitative aspects, (2) automation of spectral assignments, and (3) development of efficient computational approaches for calculation and refinement of 3-dimensional molecular structures. The goal of this project offers an improved methodology to meet directly the first challenge and to assist meeting the second challenge: optimization of spectral contrast and resolution enhancement for multi-dimensional NMR data and improvement of quantification for evaluation of molecular structures. To achieve this goal, non-linear processing methodology based on maximum likelihood spectral reconstruction is combined with selected symmetrization and other mathematical and logical operations, to enhance 2D and 3D NMR spectral contrast, and significantly facilitate identification of all crosspeaks, with subsequent improved quantification for both isolated and overlapped spectral features. this project will evaluate and extend spectral quantification using these methods, exploring whether the methodology increases the number of usable and quantifiable crosspeaks. Preliminary investigations suggest that gains of 25-50% or more can be achieved, significantly improving subsequent molecular modeling steps. Evaluation utilizes a mix of synthetic and real experimental data in a multi-part quantitation protocol, followed by experiments on four test cases: a 24-mer RNA hairpin loop, an octamer DNA mini-helix and its RNA cognate, and the small protein lysozyme.
| Borer, P N; Levy, G C (1994) Using maximum likelihood spectral deconvolution in multidimensional nuclear magnetic resonance. Methods Enzymol 239:257-88 |
