We plan to develop NMR and computational methods for structural studies of large proteins. This will include further development of multidimensional multiple resonance techniques for conformation-independent sequential assignments and for measurements of homonuclear and heteronuclear coupling constants. This may lead to a better characterization of dihedral angles, side chain orientations and orientations of the peptide planes. This will improve the possibility of identifying hydrogen bonds and may lead to higher quality structures of large proteins. These techniques require precise measurements of resonance positions. We therefore propose proton decoupling schemes that will increase resolution and sensitivity in heteronuclear multidimensional NMR spectra. Furthermore, we will extensively explore methods of processing truncated data sets, examine and apply non-linear sampling techniques and develop processing software that can handle such data sets. To make use of the increased resolution, large multidimensional data sets will have to be handled. At present, this is extremely difficult due to limitations of computing speed and disk space. Consequently, we are in a situation where high resolution and high information content obtainable with advanced NMR technology must be sacrificed due to limitations of the computational facilities. We therefore propose to purchase a fast parallel computer with a large data storage device and to port/develop NMR processing, analysis and other related software. Analysis of multidimensional NMR spectra requires extensive paging through 2D cross planes of 3D or 4D spectra. We plan to develop a mask convolution technique that will lead to a quick and robust assignment of cross peaks. This will be accompanied by development of automated assignment techniques in multidimensional NMR spectra. Finally, to direct the effort in measurements of structural parameters, we will perform simulations to estimate the impact of precise determination of homonuclear and heteronuclear vicinal coupling constants relative to and in addition to measuring cross relaxation rates precisely and in large. To improve the latter aspect we will adapt and further develop an iterative strategy proposed by Koehl and Lefevre (1990) for the separation of direct cross relaxation and spin diffusion.
Showing the most recent 10 out of 245 publications
