EFFECT OF LINEAR PREDICTION ON DISTANCE CONSTRAINTS OBTAINED FROM QUANTITATIVE-EVALUATION OF NOESY DATA IN CONJUNCTION WITH COMPLETE RELAXATION MATRIX ANALYSIS

The effects of extrapolating 2D NOESY data In the t(1) dimension using linear prediction (LP) based on singular value decomposition were exa mined to determine if data extended with this processing method accura tely reproduce quantitative distance constraints obtained from complet e relaxation matrix analysis based on NOESY data acquired with more t( 1) data points, but processed without linear prediction. NOESY data fo r the self-complementary decamer 5' dGCGAAUUCGC were collected with ei ther 256, 512 or 800 points experimentally acquired in t(1). The data were extended to 2048 points either by zero-filling or by using forwar d LP using all t(1) data points. The data were apodized, Fourier trans formed and baseline corrected and the resultant 2D NMR spectra were qu alitatively and quantitatively evaluated. Eighteen cross peaks were nu merically integrated for each set of experimental conditions and the r elative volumes for each of these cross peaks were used to estimate in terproton distances and upper and lower distance constraints via relax ation matrix analysis using the program MARDIGRAS. The reproducibility of volumes for a cross peak from a given set of experimental conditio ns was also evaluated by quantitatively assessing the same 18 cross pe aks from NOESY data collected in quadruplicate with 512 points in t(1) . The relative volumes of NOESY cross peaks do not systematically vary depending on the number of points acquired in t(1) or whether the dat a are extended with zero-filling or linear prediction based on singula r value decomposition prior to Fourier transformation, The estimated i nterproton distance, and upper and lower distance bounds, obtained fro m a relaxation matrix analysis based on the NOESY data are also insens itive to the use of LP to extend relatively smaller data sets, The res ults indicate that LP can be used to reduce the acquisition time of 2D NOESY data sets by a factor of two to three without misrepresenting t he cross peak volumes used in the determination of the three-dimension al structures of duplex DNA.