Re: correlation time and error -- May 24, 2012

Hi,

If you didn't have the temperature compensation (with either
single-scan interleaving or temperature compensation blocks), then I
would highly recommend you use Sebastien Morin's relaxation data
consistency testing analysis built into relax.  This will say if there
are problems with the data.  If this is the case, then any Rex values
you obtain should be ignored as it is likely to be fake (see the
Tjandra 96 paper about this).

For the diffusion tensor, I think in the Morin and Gagné paper you
link to (http://dx.doi.org/10.1016/j.bpj.2009.02.068), a special
script was written for this.  Sebastien might be able to say a bit
more about this.  The way you would do this is to write a custom
script which reads in the final relax results file you would like to
use, use the fix user function to un-fix the diffusion tensor while
fixing all the model-free parameters, run 500 Monte Carlo simulations,
and the save the result.  This shouldn't be too hard to write, though
you might need to read the user function documentation.

Regards,

Edward


On 24 May 2012 10:18, Romel Bobby <rbob002@xxxxxxxxxxxxxxxxx> wrote:

Hi Edward,

I have data at two field strengths with proper temperature calibration.
Unfortunately, by the time I ran the experiment we didn't have proper
temperature compensation included for the T2 experiment. However, I ran the
T2 experiment with alternating delays per t1 increment, so that the average
CPMG delay was kept more or less constant.

I used the dauvergne_protocol.py to do the S2 analysis and afterwards ran
the final_data_extraction.py script to extract the model-free parameters.

My question regarding the error for the diffusion tensor components and for
the global correlation time was in respect with a paper I read
(doi:  10.1016/j.bpj.2009.02.068). They have used relax and reported
errors.I'm inclined to do the same and could use some advice how to go about
it the correct way.

Kind regards,

Romel



On Friday, 11 May 2012, Edward d'Auvergne <edward@xxxxxxxxxxxxx> wrote:

Dear Romel,

Welcome to the relax mailing lists!  To be able to answer your
question, would you be able to give more details about your analysis?
Which scripts did you use, i.e. did you use the dauvergne_protocol.py
script?  And if not, would you be able to describe how you implemented
the full, highly iterative analysis (see my response to Fernando
Amador, http://www.mail-archive.com/relax-users@xxxxxxx/msg01158.html),
possibly attaching your hand written scripts.  And do you have data at
two or three field strengths, all with proper temperature compensation
and per-experiment temperature calibration?

As for the extraction of diffusion tensor errors, this will not be
possible with the implementation of the dauvergne_protocol.py script.
You can see the analysis implementation details in the
auto_analyses/dauvergne_protocol.py file, specifically in the
execute() method on line 525.  If you go down to the part titled
'Final run', you will see that just at the start of the Monte Carlo
simulation section that the diffusion tensor is fixed prior to the
simulations.  The reason for this is that the diffusion tensor is the
biggest contributor to the relaxation data.  Hence if it is allowed to
move, this has a large affect on the model-free parameters.  Although
it shouldn't change your model-free parameter errors too much (this
has never been investigated by the NMR field to date, as far as I
know), the result is that each simulation takes an incredibly long
time to complete.  Instead of optimising many models of 1-6
parameters, this will optimise the global model of all model-free
parameters of all models simultaneously with the diffusion tensor
parameters, i.e. a single model of hundreds to thousands of
parameters.  If you would like to perform Monte Carlo simulations on
such a global model, you will require either a few years computation
time or access to a cluster and then look into Gary Thompson's
multi-processor implementation for relax.

I hope this info helps,

Regards,

Edward



On 11 May 2012 06:35, Romel Bobby <rbob002@xxxxxxxxxxxxxxxxx> wrote:

Hi all,

First of all, I want to thank the users and developers on this mailing
list
for their constant support for people like me. This forum has helped me
with
several issues.

Recently, I have started using relax to extract model-free parameters
from
NMR spin relaxation experiments and so far I've been happy with the
software.
For my analysis, I've been using the model scripts that come with the
software and at the end, after optimisation and model selections, I've
been
extracting the results from the final run, ie after the MC simulation, by
using the script written by Michael Bieri

However, that script gives me the final values for the rotational
correlation time and parameters for the diffusion tensor among many
other.
However, from the script one cannot get the errors associated with the
correlation time (and also for the diffusion tensor parameters). Surely,
from the MC simulations it ought to be possible to get an estimation on
the
uncertainties of these values (?).
Unfortunately, I'm not very familiar with the interior of relax, or
python
for that matter, that I was able to find a solution yet.

Any kind of help/suggestion will be greatly appreciated.

Kind regards,

Romel Bobby, PhD Student
Biomolecular NMR Research Group
School of Chemical Sciences/School of Biological Sciences
The University of Auckland
+64 (09) 3737599 Ext 83157

_______________________________________________
relax (http://www.nmr-relax.com)

This is the relax-users mailing list
relax-users@xxxxxxx

To unsubscribe from this list, get a password
reminder, or change your subscription options,
visit the list information page at
https://mail.gna.org/listinfo/relax-users


--
Romel Bobby

Biomolecular NMR Research Group
School of Chemical Sciences/School of Biological Sciences
The University of Auckland
+64 (09) 3737599 Ext 83157

Re: correlation time and error

Header

Content

Related Messages