NUS for 2D experiments with TOPSPIN 3.0 and NUS for 3D 13C-separated NOESY

[Youlin Xia]

Hi, dear friends,

Is there any way to setup NUS for 2D experiments with software topspin 3.0?

3D NUS for HNCACB, HNCO, HNCA, 15N-separated NOESY and HNHA run very well
and the reconstructed data with mddnmr2.0 show ~2-time S/N improvement
compared with traditional data. But the 3D NUS 13C-separated NOESY is very
bad and the spectrum shows much smaller number of cross peaks compared
traditional NOESY. Did anyone run the 13C-separated NOESY experiment? How
are your data? Any suggestion?

Many thanks,
Youlin Xia
University of Minnesota
www.umn.edu/nmr

[Maxim A. Dubinnyi]

My attempt to acquire 3D NUS 13C-separated NOESY resulted in very
bad spectrum, the data has large noisy planes in indirect dimensions. This may
be due to autoshim function which disturbed shim values while data were acquired
(the lock level reduced by factor of two). If it is not your case, try to reduce MDD_NOISE
value from the default 0.7 to some smaller value, e.g 0.05. As for my experience,
for 15N-separated NOESY the best spectrum is obtained for MDD_NOISE=0.05,
for the default value 0.7 only the largest cross-peaks are observed.

Sincerely,
Maxim A. Dubinnyi,
Laboratory of biomolecular NMR-spectroscopy,
M.M. Shemyakin and Yu.A. Ovchinnikov Institute of bioorganic chemistry of the Russian Academy of Sciences,
Moscow, Russia.

[Vladislav Orekhov]

This may have something to do with Bruker implementation of the pulse sequence. For Varian BioPack,  13C NOESY-HSQC, NUS/MDD works fine.

Vladislav

--
You received this message because you are subscribed to the Google Groups "mddnmr" group.
To post to this group, send email to mdd...@googlegroups.com rom your registered email address.
To unsubscribe from this group, send email to mddnmr-un...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/mddnmr

[Youlin Xia]

I acquired a 3D 13C-separated NOESY (NOESYHSQCETGPSI3D) with traditional method. The spectrum looks good. The parameters are d1=1, ns=8, d8=0.2, TD=(70, 160, 2k).

 

Then I acquired three corresponding NUS spectra with sparse rate=30% & J=(0,0,0), and T2=(0.02, 0.02, 1.0), (0.05, 0.10, 1.0), or (0.02, 0.05, 1.0). I used mddnmr2.0 to process the data with parameters METHOD=’MDD’, SRSIZE=0.1, NCOMP=25 to 600, NITER=50 to 100, OVLP=3 MDD_NOISE=0.7, lambda=0.01, CT_SP=’nnn’, CEX=’nnn’.

 

To process other 3D data including 15N-separated NOESY, above mddnmr parameters seem ok. But I’ll try to reduce MDD_NOISE as Maxim suggested to test.

[Vladislav Orekhov]

On 2011-12-13 17.14, Youlin Xia wrote:

I acquired a 3D 13C-separated NOESY (NOESYHSQCETGPSI3D) with traditional method. The spectrum looks good. The parameters are d1=1, ns=8, d8=0.2, TD=(70, 160, 2k).

 

Then I acquired three corresponding NUS spectra with sparse rate=30% & J=(0,0,0), and T2=(0.02, 0.02, 1.0), (0.05, 0.10, 1.0), or (0.02, 0.05, 1.0). I used mddnmr2.0 to process the data with parameters METHOD=’MDD’, SRSIZE=0.1, NCOMP=25 to 600,

Calculations with NCOMP=600 should take forever. Are you sure about the number? Anyway, I suggest 25-40.

NITER=50 to 100,

The number of iterations may be be too little and is good only for the first look at the spectrum. For the final calculations, try 200-500. It will lengthen the calculations, of cause.

[Youlin Xia]

Dear Dr. Orekhov,

 

It’s very nice to have the chance to discuss mddnmr with you. The enclosed is a table to compare S/N for traditional and NUS HNCACB and HNCA. I randomly chose three peaks. And I changed parameters NCOMP and NITER. It seems to me NCOMP=600 and NITER=100 are good, which gives me more than 2-time S/N improvement compared with traditional way. Increasing NCOMP can significantly decrease noise level.  Please note that the factor 1.825 is from sqrt(15/4.5). This is because the traditional HNCACB took 15 hours and NUS HNCACB only took 4.5 hours, and the S/N of NUS spectrum should be multiplied by 1.825 if it took 15 hours instead of 4.5 hours.

 

What else parameters should I change? I may play with the parameter MDD_NOISE.