Heya Giacomo Quilici,
I don't really know a whole lot about NUS in Topspin, but as I understand, you would probably have had to reconstruct the data (presumably within Topspin), to get a much higher resolution spectrum than normal sampling would. When you compare your NUS spectrum
to the standard spectrum - are you comparing the same number of increments (i.e. 2k x 128 x 128) or is your NUS spectrum higher (2k x 1k x 1k)? A simple way to check this is just look at the size of the file of 3rrr between your NUS sample and the conventional
sampled file.
If this is the issue, the problem is largely hardware related as your computer has to try and read gigabytes of data in from the hard drive to display a single plane and even modern computers aren't the best at this.
Possible work arounds:
1) Use contour files rather than raw processed data. This was mentioned a few weeks ago by one of the developers. Basically, you get the spectrum to an appropriate S/N ratio that you want to work at, and you can then record a reduced data file that is much
quicker to display the contours than recalculating them every time you move through a 2D plane. Only issue is that you can't increase or decrease contour levels as you go (you need to recalculate the contour file or turn it off). To use them you get to the
dialogue box in the spectra window at the bottom ("contour files"), then create a contour file for each of your spectra. Be sure to select the two dimensions you are going to be looking at (i.e. usually 1H and 13C for the HNCA, etc). This will take a long
time especially if your 3D data file is huge. You then set 'use contour files' as 'yes' in the spectra window. Check out
http://www2.ccpn.ac.uk/documentation/analysis/popups/CreateContourFilePopup.html
and the archive of this discussion board for more information. As I said, I think it was only a few weeks ago this was discussed.
2) Use something like nmrPipe and order your data in the planes you expect to view the spectrum in. This is more involved, and I suspect that you wouldn't bother with this if you are going with the contour file approach, but if you have your NMR data stored
as f3f2 planes (i.e. 1H15N usually for Bruker stuff at least), you need to read each plane to plot a new 2D trace in f3f1 (1H13C). You significantly reduce the work the computer has to do if you store 1H13C planes as it only has to find the right one and display
it, not read each plane. If you don't quite get what I mean, it is probably more hassle than the speed improvement would be. There is no improvement using this method if you use contour files.
If the size of your processed data is not any bigger than the conventionally sampled stuff, perhaps there is another issue here, but you would need advice from the developers on that one.
All the best,
Tom
On 10/10/13 7:09 PM, Giacomo Quilici wrote:
[log in to unmask]">
Good morning,
I have recently acquired 3D spectra (HNCa and HNCO) on a Bruker spectrometer in Non Uniform Sampling (NUS) way.
I loaded them on a CCPNMR analysis (2.1) project. The spectra visualization is ok, but it is extremely slow (almost impossible to manage) to navigate through the spectra with respect to same spectra acquired without NUS.
I loaded them in "bruker" format selecting (as usual) the entire processing folder. I have already checked if the dimension of 3irr-3rrr files is greater with respect to the homologues spectra aqcuired in standard way, but this is not the case.
any suggestions?
best regards
Giacomo Quilici
--
Thomas J Carruthers
Structural Biology and Biophysics
Research School of Chemistry
Australian National University
Building 137-1.17
Sullivans Creek Road
CANBERRA ACT 0200 AUSTRALIA
E:
[log in to unmask]
P: (+61) 2 612 56506
