Re: back calculation of relaxation data

[Christina Möller]

Dear Edward,

thank you so much for your recommendations and the really helpful wiki
article about the CGS and SI unit system. Now I understand the equation
differences and successfully back-calculated my relaxation data.

Furthermore, I used the generate_ri.py script and the back-calculated
values are close to the experimental data (as expected). I have to admit
that I copied the motion parameters (s2,sf2,s2s,tf, etc) and the
associated model manually into the script and executed the script
several times for different residues. I am pretty sure that it is
possible to include one or several lists in order to automatically
update these parameters for each residue and run this script only one
time. Unfortunately, I was not able not find out how this works. Could
you give me an idea?

Thanks,
Christina



Am 11/05/2015 um 03:53 PM schrieb Edward d'Auvergne:

Hi Christina,

Please see below:

On 5 November 2015 at 14:05, Christina Möller  wrote:

Dear Edward and relax-users,

I am sorry to ask you again for some advice to back calculate the relaxation
data from the model parameter values. Here is what I tried so far:
For the back calculation I loaded the final results determined by the
automated analysis using the dauvergne_protocol.py into the relax gui. Then
I selected the back_calc tool from user functions -> relax_data.
After starting the back calculation without any entry
relax> relax_data.back_calc(ri_id=None, ri_type=None, frq=None)
I get the following error message:
RelaxWarning: comparison to `None` will result in an elementwise object
comparison in the future.

This is a warning produced by more recent versions of the numpy Python
package.  It is harmless message about changes which will occur in
numpy, and it is safe to ignore.  Note that I have fixed this in the
source code repository and the fix will appear in the future when I
release relax 4.0.1 (it is not present in relax 4.0.0 at
http://wiki.nmr-relax.com/Relax_4.0.0 ).  This was first reported at:

 [relax-users] RelaxWarning: comparison to `None` will result in an
elementwise object comparison in the future (
http://thread.gmane.org/gmane.science.nmr.relax.user/1839 ).

And the solution at:

 [relax-users] Re: RelaxWarning: comparison to `None` will result
in an elementwise object comparison in the future (
http://thread.gmane.org/gmane.science.nmr.relax.user/1844 ).



If I put the frequency in
relax> relax_data.back_calc(ri_id=None, ri_type=None, frq=599468076.0)
I get another error message
Traceback (most recent call last):
  File
"/opt/scisoft64/usr/lib/python2.7/site-packages/relax/gui/interpreter.py",
line 306, in run
fn(*args, **kwds)
  File
"/opt/scisoft64/usr/lib/python2.7/site-packages/relax/pipe_control/relax_data.py",
line 132, in back_calc
spin.ri_data_bc[ri_id] = api.back_calc_ri(spin_index=spin_index,
ri_id=ri_id, ri_type=ri_types[ri_id], frq=frqs[ri_id])
KeyError: 'R1_600'

Can you give me a hint what I am doing wrong?

I would suggest using the sample_scripts/model_free/generate_ri.py
sample script.  The relax_data.back_calc user function requires that
the relaxation data ID, the relaxation type, and frequency arguments
be supplied to correctly calculate the desired data (see
http://www.nmr-relax.com/manual/relax_data_back_calc.html ).  Without
this information, relax cannot determine what you would like to do.
Using the sample script would be the easiest option.



I know that the relaxation data can also be back calculated using the
equations 7.3a - 7.8 in the relax manual (PDF), although it is not that
trivial for an ellipsoid model because I need to determine the weights. Is
there a reason why you define c=1/3(ωH*∆σ)^2 (eq. 7.5) instead of
c=2/15(ωN*∆σ)^2 like it can be found in literature?

These two factors come from using CGS vs. SI units.  I have mentioned
this issue before (see
http://thread.gmane.org/gmane.science.nmr.relax.devel/1023/focus=1034
).  I have also started a stub wiki article to help explain these
differences - http://wiki.nmr-relax.com/CGS_versus_SI . In SI units
the correct unit (squared) is (ω·∆σ)^2 / 3, and SI units are used
throughout relax.  This has made some parts of relax more difficult to
implement as there are many equations in the literature in CGS units,
and the authors will often not identify this older system.  Some
stubborn people just prefer the magnetic constant to be 1 rather than
µ0/(4π), and will not shift from the CGS metric system to the SI
metric system.

Regards,

Edward




--

Christina Möller
ICS-6 / Structural Biochemistry
Forschungszentrum Jülich
D-52425 Jülich
GERMANY

E-mail: c.moel...@fz-juelich.de
Tel.: +49-2461-619387
Fax +49-2461-619387





Forschungszentrum Juelich GmbH
52425 

Re: back calculation of relaxation data

[Edward d'Auvergne]

On 5 November 2015 at 15:53, Edward d'Auvergne  wrote:
> On 5 November 2015 at 14:05, Christina Möller  wrote:
>> I know that the relaxation data can also be back calculated using the
>> equations 7.3a - 7.8 in the relax manual (PDF), although it is not that
>> trivial for an ellipsoid model because I need to determine the weights. Is
>> there a reason why you define c=1/3(ωH*∆σ)^2 (eq. 7.5) instead of
>> c=2/15(ωN*∆σ)^2 like it can be found in literature?
>
> These two factors come from using CGS vs. SI units.  I have mentioned
> this issue before (see
> http://thread.gmane.org/gmane.science.nmr.relax.devel/1023/focus=1034
> ).  I have also started a stub wiki article to help explain these
> differences - http://wiki.nmr-relax.com/CGS_versus_SI . In SI units
> the correct unit (squared) is (ω·∆σ)^2 / 3, and SI units are used
> throughout relax.  This has made some parts of relax more difficult to
> implement as there are many equations in the literature in CGS units,
> and the authors will often not identify this older system.  Some
> stubborn people just prefer the magnetic constant to be 1 rather than
> µ0/(4π), and will not shift from the CGS metric system to the SI
> metric system.

Hi Christina,

The wiki article http://wiki.nmr-relax.com/CGS_versus_SI is now
somewhat improved.  However it doesn't cover all of the NMR constants
yet, as that will be a lot of work to derive.  Anyway, I hope that
helps a little to understand the equation differences seen in the NMR
literature.

Regards,

Edward

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

This is the relax-users mailing list
relax-users@gna.org

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

Re: back calculation of relaxation data

[Edward d'Auvergne]

Hi Christina,

Please see below:

On 5 November 2015 at 14:05, Christina Möller  wrote:
> Dear Edward and relax-users,
>
> I am sorry to ask you again for some advice to back calculate the relaxation
> data from the model parameter values. Here is what I tried so far:
> For the back calculation I loaded the final results determined by the
> automated analysis using the dauvergne_protocol.py into the relax gui. Then
> I selected the back_calc tool from user functions -> relax_data.
> After starting the back calculation without any entry
> relax> relax_data.back_calc(ri_id=None, ri_type=None, frq=None)
> I get the following error message:
> RelaxWarning: comparison to `None` will result in an elementwise object
> comparison in the future.

This is a warning produced by more recent versions of the numpy Python
package.  It is harmless message about changes which will occur in
numpy, and it is safe to ignore.  Note that I have fixed this in the
source code repository and the fix will appear in the future when I
release relax 4.0.1 (it is not present in relax 4.0.0 at
http://wiki.nmr-relax.com/Relax_4.0.0 ).  This was first reported at:

[relax-users] RelaxWarning: comparison to `None` will result in an
elementwise object comparison in the future (
http://thread.gmane.org/gmane.science.nmr.relax.user/1839 ).

And the solution at:

[relax-users] Re: RelaxWarning: comparison to `None` will result
in an elementwise object comparison in the future (
http://thread.gmane.org/gmane.science.nmr.relax.user/1844 ).


> If I put the frequency in
> relax> relax_data.back_calc(ri_id=None, ri_type=None, frq=599468076.0)
> I get another error message
> Traceback (most recent call last):
>  File
> "/opt/scisoft64/usr/lib/python2.7/site-packages/relax/gui/interpreter.py",
> line 306, in run
>fn(*args, **kwds)
>  File
> "/opt/scisoft64/usr/lib/python2.7/site-packages/relax/pipe_control/relax_data.py",
> line 132, in back_calc
>spin.ri_data_bc[ri_id] = api.back_calc_ri(spin_index=spin_index,
> ri_id=ri_id, ri_type=ri_types[ri_id], frq=frqs[ri_id])
> KeyError: 'R1_600'
>
> Can you give me a hint what I am doing wrong?

I would suggest using the sample_scripts/model_free/generate_ri.py
sample script.  The relax_data.back_calc user function requires that
the relaxation data ID, the relaxation type, and frequency arguments
be supplied to correctly calculate the desired data (see
http://www.nmr-relax.com/manual/relax_data_back_calc.html ).  Without
this information, relax cannot determine what you would like to do.
Using the sample script would be the easiest option.


> I know that the relaxation data can also be back calculated using the
> equations 7.3a - 7.8 in the relax manual (PDF), although it is not that
> trivial for an ellipsoid model because I need to determine the weights. Is
> there a reason why you define c=1/3(ωH*∆σ)^2 (eq. 7.5) instead of
> c=2/15(ωN*∆σ)^2 like it can be found in literature?

These two factors come from using CGS vs. SI units.  I have mentioned
this issue before (see
http://thread.gmane.org/gmane.science.nmr.relax.devel/1023/focus=1034
).  I have also started a stub wiki article to help explain these
differences - http://wiki.nmr-relax.com/CGS_versus_SI . In SI units
the correct unit (squared) is (ω·∆σ)^2 / 3, and SI units are used
throughout relax.  This has made some parts of relax more difficult to
implement as there are many equations in the literature in CGS units,
and the authors will often not identify this older system.  Some
stubborn people just prefer the magnetic constant to be 1 rather than
µ0/(4π), and will not shift from the CGS metric system to the SI
metric system.

Regards,

Edward

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

This is the relax-users mailing list
relax-users@gna.org

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