Hi,
I have also encountered this scaling problem, which is indeed attributed
to different spectra being processed with different values of the
NC_proc parameter. I think the idea is that the Bruker data is stored
in integer format and has to be scaled to avoid the possibility of
integer overflow during operations such as phase correction. I could be
wrong, but my understanding is that the parameter works in the opposite
sense to that described by Horst i.e. NC_proc=2 means that the spectrum
has been scaled DOWN by a factor of 4.
Although it would be nice if this scaling could be accounted for in
Analysis, in the meantime the problem can be avoided by processing the
data in Topspin in the following way:
1. Process all your spectra as normal using xfb (or multixfb)
2. For each spectrum use the command "dpp" and note the value of
NC_proc (or alternatively use grep to search the procs files: i.e. %grep
NC_proc <path to your data>/*/pdata/1/procs)
3. Note the largest value and then reprocess the data using the command
"xfb nc_proc <your largest NC_proc value>" i.e. if you had three spectra
for which the values of NC_proc were -3, -1 and 2 your would use "xfb
nc_proc 2"
4. Use dpp (or grep) to verify that all spectra now have the same value
of NC_proc.
The data can now be used with Analysis.
If you have many spectra to process you can make a modified version of
the Topspin multixfb au program by replacing the XFB command with
XCMD("xfb nc_proc <your largest NC_proc value>"). This would need
updating with the relevant value for NC_proc each time it was used.
With a few extra lines you can have the program prompt you for this
parameter.
HTH,
Pat Edwards.
-----Original Message-----
From: CcpNmr software mailing list [mailto:[log in to unmask]] On
Behalf Of Wayne Boucher
Sent: Wednesday, 13 February 2008 9:38 p.m.
To: [log in to unmask]
Subject: Re: problem with T1 series
Thanks, that's great. (I guess there's probably a Bruker manual that
probably specifies all of this but I don't have a copy of it.) I'll
have to see how/if we can work that into the current Analysis. (It will
definitely be in the new version.)
Wayne
On Wed, 13 Feb 2008, Dr Horst Schirra wrote:
> Hi Wayne and Mark,
>
> the parameter you want is called NC_proc and will be in the procs
> processing parameter file for each spectrum.
>
> The parameter is a log2 scale, i.e.:
>
> NC_proc=2 spectrum has been scaled up by factor of 4
> NC_proc=1 spectrum has been scaled up by factor of 2
> NC_proc=0 scaling of spectrum intensity unchanged
> NC_proc=-1 spectrum has been scaled down by factor of 2 NC_proc=-2
> spectrum has been scaled down by factor of 4
>
> etc.
>
> In my relaxation measurements, I typically encounter NC_proc values
> between -2 and -6, just as an indication...
>
> The true intensity of any data point is then
> int_corr= bruker_raw_int * (2 ^ NC_proc)
>
> Cheers,
>
> Horst Joachim Schirra
>
> PS: Can you tell, I had to fiddle with that *a lot*? ;)
> ----------------------------------------------------------------
> / Dr. sc. nat. Horst Joachim Schirra Phone: (+61)7/3346-2021 /
> / Queensland Smart State Fellow Fax: (+61)7/3346-2101 /
> / Institute for Molecular Bioscience /
> / University of Queensland email: [log in to unmask] /
> / Brisbane QLD 4072, Australia http://www.uq.edu.au/~uqhschir /
> ----------------------------------------------------------------
>
>
>
> ----- Original Message -----
> From: Wayne Boucher <[log in to unmask]>
> Date: Tuesday, February 12, 2008 9:03 pm
> Subject: Re: problem with T1 series
> > Hello,
> >
> > This is a (relatively recently) known problem. Bruker stores
> > everythingas ints, not floats, so quite sensibly has a multiplier
> > specified (which we discovered when someone else had a problem
> > similar to the below). This requires us to add a scale to the C
> > world, passed down from the Python world, so that the data comes
> > back correctly. For contouring there
> > is a fix already available (setting the spectrum scale, although
> > I'll have
> > to remind myself where this scale is hiding in the Bruker files).
> > Thisfix is not available for the rates analysis but I guess could
> > be (if Tim
> > thinks it makes sense).
> >
> > Wayne
> >
> > On Tue, 12 Feb 2008, Mark Pfuhl wrote:
> >
> > > I tried to extract T1 values from a standard T1 experiment. The
> > spectrum was
> > > recorded and processed with Topspin and loaded straight into
> > analysis. In
> > > Topspin I got the expected decay but in analysis I have a
> > sawtooth pattern
> > > that has no resemblance to the original data.
> > > The sawtooth appearance suggests that the exponent of the
> > intensities is
> > > lost somehow in loading it into analysis.
> > > Is there something wrong or do I need to load the spectra in a
> > specific way
> > > if I want to maintain correct intensities?
> > > thanks,
> > > mark
> > >
> >
>
|