The data shown here is a phase-sensitive NOESY spectrum of codeine, displayed with the direct dimension horizontal.
Note the "t1 noise" at the positions of the 2 methyls.
The noise causing the streaks exists in only 2 slices per methyl peak. The offending slices can be edited to give a much nicer plot.
Because NUTS displays only horizontal slices, first transpose the data (TD) so the slices containing streaks can be accessed.
A stacked plot shows why we see streaks in the contour plot. The noise level in 2 slices is much greater than in the rest of the data. We don't want to zero out these slices, as they contain useful crosspeaks.
Use the View routine (VW) to step through slices and the Slice command (SL) to display a selected slice. Editing of selected slices is simpler to do in arrayed mode, because it is not necessary to save the data after editing each slice. (If not in arrayed mode, the S2 command is used to save after each modification - if you don't save, when you display a different slice, your changes will be lost.) As you look through the data, there may be slices with distorted baselines, which can be corrected with FB or QB.
To remove the streaks, we reduce the amplitude of all points in the 2 offending slices, and the noise level becomes more comparable to that of the rest of the data. This was done using the NUTS divide command, reducing the amplitude of all data points in the 2 selected slices by a factor of 4. This is sufficient to eliminate the streak from the contour plot without affecting the crosspeak.
In some cases, a single slice of a 2D data set may be corrupted. It may be possible to salvage the data by removing the offending slice. This can be done by zeroing all points in that slice (with ZE or, more conveniently, the View subcommand Z) or, usually better, substituting the adjacent slice for the one that is corrupted.
Last updated: 3/17/02