URL:
<http://gna.org/support/?3154>
Summary: Implementation of Baldwin (2014) B14 model - 2-site
exact solution model for all time scales
Project: relax
Submitted by: tlinnet
Submitted on: Tue 29 Apr 2014 10:25:52 PM UTC
Category: Feature request
Priority: 5 - Normal
Severity: 3 - Normal
Status: None
Assigned to: tlinnet
Originator Email:
Open/Closed: Open
Discussion Lock: Any
Operating System: None
_______________________________________________________
Details:
This is a feature request for the implementation of Baldwin (2014) B14 model -
2-site exact solution model for all time scales.
http://dx.doi.org/10.1016/j.jmr.2014.02.023
"An exact solution for R2,eff in CPMG experiments in the case of two site
chemical exchange"
Andrew J. Baldwin
Journal of Magnetic Resonance
Main feature:
In practise, significant deviations from the Carver Richards equation can be
incurred if pB>1%. Incorporation of the correction term into equation (article
equation 50), results in an improved description of the CPMG experiment over
the Carver Richards equation.
##################################################
##### Comments from the Author, revised by Troels.
The formula is quite similar to the Carver Richards, so it'll be a bit slower.
It terms of raw speed, I've found it about 100 times faster than a numerical
solution, and about 3x slower than Carver Richards.
That fits roughly with the number of extra function calls.
Note that using arc-cos, rather than square roots to get the prefactors for
the Eigenvalues in the first few lines speeds things up a little. This little
timesaver would work also in your implementation of the Carver Richard's
formula.
The advantage of this code will be that you'll always get the right answer
provided you've got 2-site exchange, in-phase magnetisation and on-resonance
pulses. With Mieboom and Carver Richard's equations, on occasion, you'll get
the wrong answer (in this scenario).
I wonder why it would ever make sense to use either of these?
If you ever saw a better fit when using either, in terms of chi2, it would be
fake, as they all reduce into each other, so the limiting cases are utterly
redundant?
Following on from that, given that in any real experiment (apart from the rare
case where you explicit decouple during the CPMG period) you'll have elements
of scalar coupling, pulses will rarely be on-resonance and you'll have
additional relaxation effects like spin-flips.
So what is the benefit of ever using a formula over a numerical solution to
fit CPMG data that incorporates this? The applications in recent years from
the Kay group for getting things like excited state structures, and the bench
marks for the experiments themselves all require inclusion of all the
physics. You cannot do this with formulas.
You might find it interesting to read e.g. the comments on this thread:
https://plus.google.com/s/cpmg%20glove
It seems the future should instead go for stop the use of formulas and instead
use numerically with accurate experimentally determined spin flip rates, R1s,
carrier positions and chemical shifts.
It's a serious point: people trying to get meaning from their data should
really not use any of these formulas.
As a first guess, fair enough.
The parameters might not be wrong. But any user, especially one that doesn't
know much about the details, stands a reasonable chance of getting a wrong
answer.
This can be avoided by insistence on doing things rigorously.
You are very welcome to incorporate this formula.
I have a version in C that is again much faster that you are welcome to get.
I think it would help if you nudge them in a more numerical direction and
force them to measure a full set of e.g. spin flip rates.
There are a number of Kay papers you can cite that would very much support
this.
All best wishes,
Andy
_______________________________________________________
Reply to this item at:
<http://gna.org/support/?3154>
_______________________________________________
Message sent via/by Gna!
http://gna.org/
[sr #3154] Implementation of Baldwin (2014) B14 model - 2-site exact solution model for all time scales