Subject: Re: [nmr_sparky] how sparky deal with 1H-15N relaxation data?
From: Josh Ward
Date: Apr 4, 2006

Previous: 101 Next: 103


I wish I understood this well enough to give a truly outstanding answer,
but this is as close as I can get. You are correct that Sparky
calculates T1 and T2 data. By definition, R = 1/T, so R is easy to
calculate. The relative (percentage) error in R is the same as the
relative (percent) error in T (see any basic statistics or analytical
chemistry textbook on how to calculate the propagation of errors in
equations).

R = 1/T
Rerr/R = Terr/T -- Rerr = (Terr/T)*R = (Terr/T)*(1/T) = Terr/T**2 (One
of those expressions should work ;-) )

As for how Sparky calculates this from a Monte Carlo simulation. Random
data sets are generated by calculating gaussian distributions around the
data points, and then the standard deviation of these datasets is
determined. You can adjust the number of trials made in the RH setup.
Default is 5, I usually increase this to 500 or so until several rounds
of RH fitting give me a reasonably similar error estimate.

And finally, for the error estimate in the NOE measurements, the
literature states that the noise should be equal to the square root of
the baseline noise. You can get the noise estimate for each spectrum
from the Spectrum information dialog menu ( View - More - Spectrum...
st ). Here I calculate noise as a median of 500 to 1000 randomly sample
data heights. I havent gotten into the extact algorithms used by
Sparky. Maybe somebody more familiar with the coding can help out there.

A couple of references where I uncovered most of this mystery:

Farrow et al Biochemistry (1994) 33:5984
Palmer et al JACS (1991) 113(12):4371


Happy relaxing!

Josh

jleeweir wrote:

hi,everyone:

Now Im working on dynamics of protein and have many questions!

First,sparky can deal with T1,T2 with a seris of 2D spectra and get
error without duplicating several experiments.How sparky do that? Is
the error of R1 or R2 equal wih T1 and T2?

Second,I dont konw how to deal with noe via sparky? How to get its
error?

Thanks to all!






Yahoo! Groups Links









--
Josh Ward
Graduate Research Assistant
Purdue University
Department of Medicinal Chemistry and Molecular Pharamacology
Lily Hall of Life Sciences
Phone: (765) 494-2191