AMMRL:
I want to thank many of you for your very quick response to my question.
Below is the request I sent and then the responses. Thanks to Steve, John,
Roger, Rolf and Michael for your help.
Best regards,
Pat Hays
===========
My request:
I am using a Varian Mercury 400 and Inova 600 for quantitative experiments
and would like to run samples at differing concentrations of a single compound,
measure the integral of one peak in each experiment and determine how well
this approach works for quantitation (no reference internal standard). I also
want to run arrayed experiments and integrate peaks and determine how the
populations are influenced by the experimental parameters. I know that there are
many integration commands available (dpirn, dpir, liamp, dli) but some, maybe
all of these, have referencing variables that could interject problems.
Has anyone done this kind of work and does anyone know which command(s) are
used so that you can integrate a peak(s) and the integral value you get is
truly proportional to the nuclei population? Are there parameters or variables
that must be set (or deactivated) to do this kind of work? I do know that
receiver gain must remain constant from sample to sample and experiment to
experiment.
Thanks for your help.
Patrick Hays
-----------
Responses to the above request:
This may not quite answer your question, but my understanding is that a T1
experiment does just you are trying to do. A T1 experiment uses intensity not
area but the two should correlate. You do need to set aig='ai' (absolute
intensity group to absolute intensity) to prevent it from scaling the peaks (I
assume that scaling the peak heights also scales the areas); aig is normally
set to nm (normalized mode). The other obvious factor that needs to remain
constant between experiments is the number of scans. Hope this helps.
----------
You should make sure ins, insref, and fn are the same, and set insref to 0 or
1. The following is from the description of insref in the Command and
Parameter Reference:
"The reported integral values will be (integral value)*ins/insref/fn. If
insref is 'n', the sum of all integrals will be ins. The 'n' mode is the
equivalent of the normalized integral mode. If insref is zero or not defined, the
reported integrals will be (integral value)*ins/fn."
Then use dli, dpir, or pir to display the integrals. You can also find them
in the array variable liamp (even-numbered elements only).
Good luck,
-------------
If you use an arrayed experiment, then the integrals scale correctly between
each spectrum. You just want to make sure the baseline is flat ( see 'dc' and
'bc' commands). This is also an easy way to compare integrals of different
experiments - create a pseudo-array. For example you can collect a spectrum
each in exp1 and exp2 with the same parameters, then use the "add('new') " or
"spadd('new') " commands to move these to the special buffer exp5. They should
now behave as an array. See add, spadd in the command manual.
Regards,
--------------
I calibrated my integrals to mM when we installed the instrument, and it's
been very handy. The easy way is to put the cursor in your calibration peak and
use the "Set Int" menu button. Save the parameters back in stdpar if you
want the standard "Setup" spectra to be calibrated in mM (at least for a single
90 on a well-tuned probe, or whatever your conditions were when you
calibrated).
Look up the parameters ins and insref in the 'Command and Parameter
Reference'. It indicates you set 'insref='n' if you want relative integrals. But
play with it aggressively - I recall getting into situations where the
integrals displayed by mark and the integrals displayed on plots don't agree ... there
may have been a difference between setting it before I took an FID and
changing it after.
-----------------
the key thing to remember in VNMR 6.1 / VnmrJ integration is that "is" is
used purely to scale the displayed / plotted intergral trace, while "ins" is used
to scale the numeric integral values as reported by "dli" "pir", etc.; as
long as "insref" is either 0 or not defined (use "destroy('insref')" to eliminate
the parameter if you want to, but setting it to 0 should be good enough), the
integrals reported by the above commands (NOT of course the "n" version of
these commands) ARE absolute. They are independent of "nt" / "ct" and
independent of "fn" because we divide by "ct" and we divide by "fn". Note that setting
"insref='n'" is NOT the same as "insref=0"!!! the attached file
"faq/vnmr_integration" should explain this in more detail - if you have further questions,
please don't hesitate to ask.
(The "faq/vnmr_integration" is found on the Varian Website User Pages under
"Software FAQ: Integration in VNMR 5.3 and 6.1 (2004-12-21)". The responder
sent all of this information which is fairly long and cover not only 1D but
2D integration. I chose not to include it in this summary- Pat).
Received on Tue Jan 18 2005 - 09:04:43 MST