Difference between revisions of "Model-free analysis single field"

From relax wiki
Jump to navigation Jump to search
m (Switch to the {{relax developer link}} template to remove dead Gna! links.)
 
(38 intermediate revisions by 2 users not shown)
Line 1: Line 1:
 +
__TOC__
 +
 
== Reference ==
 
== Reference ==
http://thread.gmane.org/gmane.science.nmr.relax.user/1552
+
https://sourceforge.net/p/nmr-relax/mailman/message/36569608/
 
 
== Text ==
 
Hi Ivan,
 
 
 
To continue:
 
 
 
'''On another note, I wonder if it is possible to modify the nmr-relax programme so that I can do model-free analysis with data from only one field strength? Alternatively, do you know of any programme (that can be installed on Windows) that can do such analysis? My work focused mainly on small molecule and ligand-based NMR and I have only just very recently started looking in to protein dynamics so I am still experimentinng different software and data treatment etc. '''
 
 
 
Firstly, the subject of single field strength data has been discussed numerous times on this mailing list.  I would recommend you read my
 
previous responses to questions relating to single field strength
 
data, and look the other messages in those threads.  You will find
 
these discussions quite informative and highly detailed:
 
  
#- Martin Ballaschk: http://thread.gmane.org/gmane.science.nmr.relax.user/1409/focus=1438
+
== The Question ==
#- Shantanu Bhattacharyya: http://thread.gmane.org/gmane.science.nmr.relax.user/1367/focus=1369
 
#- Mengjun Xue: http://thread.gmane.org/gmane.science.nmr.relax.user/1276/focus=1277
 
#- Fernando Amador: http://article.gmane.org/gmane.science.nmr.relax.user/84
 
#- Shantanu Bhattacharyya: http://thread.gmane.org/gmane.science.nmr.relax.user/1086/focus=1087
 
#- Dhanasekaran Muthu: http://thread.gmane.org/gmane.science.nmr.relax.user/1152/focus=1153
 
#- Vitaly Vostrikov: http://thread.gmane.org/gmane.science.nmr.relax.user/1147/focus=1150
 
#- Aldino Viegas: http://thread.gmane.org/gmane.science.nmr.relax.user/1127/focus=1128
 
#- Pierre-Yves Savard: http://thread.gmane.org/gmane.science.nmr.relax.user/724/focus=725
 
#- Keith Constantine: http://thread.gmane.org/gmane.science.nmr.relax.user/513/focus=517
 
#- Clare-Louise Evans: http://thread.gmane.org/gmane.science.nmr.relax.user/326/focus=332
 
#- Hongyan Li: http://thread.gmane.org/gmane.science.nmr.relax.devel/694/focus=701
 
  
These will have lots of additional information.
+
'''Is possible to use relax to perform a model-free analysis using only relaxation data collected at a single magnetic field strength?'''
This is just a selection of possibly the most useful messages.
 
  
 +
The answer is yes!
  
You will soon see that this is a complicated topicNote that relax
+
The subject of single field strength data has been discussed numerous times on this mailing listIt is recommended to read previous responses to questions relating to single field strength data, and look the other messages in those threadsYou will find these discussions quite informative and highly detailed:
is capable of performing 100% of the functionality of Modelfree4 (with
 
or without the Fast-Modelfree GUI interface), Dasha, Tensor2, and
 
DYNAMICSIf you play with the optimisation settings you can even
 
find identical results to within machine precision - relax can mimic
 
these other softwares.
 
  
The key is that the full analysis protocol is rather complicated -
+
{| class="wikitable sortable"
many people don't understand this - and that these softwares do not
+
|-
implement the full iterative protocol. Therefore you either have to
+
! Person
perform it manually or write a script to perform all of the steps.
+
! Message date
The protocol is described in the relax manual in figure 7.2
+
! Message ID
(http://www.nmr-relax.com/manual/diffusion_seeded_paradigm.html). In
+
|-
summary:
+
| [https://sourceforge.net/p/nmr-relax/mailman/message/36569855/ Danyun Zeng]
 +
| {{#time: Y-m-d H:i:s|Wed, 28 Jan 2015 10:40:23 +0100}}
 +
| <small><CAED9pY-m5YEjN6KUX88s_yXLpY1BCudUaVvWWmC0U6WkteDnPA@mail.gmail.com></small>
 +
|-
 +
| [https://sourceforge.net/p/nmr-relax/mailman/message/36569841/ Sean Leo Moro]
 +
| {{#time: Y-m-d H:i:s|Fri, 12 Dec 2014 09:24:51 +0100}}
 +
| <small><CAED9pY9eAa7B+pC2+RU_pTXWR14fgtiMAiJdQ7rKw+aY6mALaA@mail.gmail.com></small>
 +
|-
 +
| [https://sourceforge.net/p/nmr-relax/mailman/message/36569491/ Navratna Vajpai]
 +
| {{#time: Y-m-d H:i:s|Tue, 30 Apr 2013 17:32:21 +0200}}
 +
| <small><CAED9pY-JcYpQtVpMe2dQf_iLvztQzVue=2d0mEQzYYnaYcRpnA@mail.gmail.com></small>
 +
|-
 +
| [https://sourceforge.net/p/nmr-relax/mailman/message/36569422/ Shantanu Bhattacharyya]
 +
| {{#time: Y-m-d H:i:s|Mon, 7 Jan 2013 11:29:28 +0100}}
 +
| <small><CAED9pY_aUzPH0OkZhTc2H=trV1VZ074XMcfT8SHYmKH5Bq6T3Q@mail.gmail.com></small>
 +
|-
 +
| [https://sourceforge.net/p/nmr-relax/mailman/message/36569341/ Mengjun Xue]
 +
| {{#time: Y-m-d H:i:s|Thu, 6 Sep 2012 14:14:19 +0200}}
 +
| <small><CAED9pY_R8s31My1w19uh5ss7Cjhh6WJLEs62sUUHsF8SxeEc7g@mail.gmail.com></small>
 +
|-
 +
| [https://sourceforge.net/p/nmr-relax/mailman/message/36569242/ Fernando Amador]
 +
| {{#time: Y-m-d H:i:s|Fri, 11 May 2012 09:34:27 +0200}}
 +
| <small><CAED9pY-gtfP__28VpkKgGgRVPAST0S0_Q+Hi94Zre0N=NfVLWg@mail.gmail.com></small>
 +
|-
 +
| [https://sourceforge.net/p/nmr-relax/mailman/message/36569074/ Dhanasekaran Muthu]
 +
| {{#time: Y-m-d H:i:s|Tue, 31 Jan 2012 15:45:22 +0100}}
 +
| <small><CAED9pY8UArGMiEf-3_fCjQrBJYPecu+BCGqkqtcSXwSix4e5cQ@mail.gmail.com></small>
 +
|-
 +
| [https://sourceforge.net/p/nmr-relax/mailman/message/36569071/ Vitaly Vostrikov]
 +
| {{#time: Y-m-d H:i:s|Mon, 23 Jan 2012 19:05:12 +0100}}
 +
| <small><CAED9pY8SGxe_rF6RKWmXJwyXjrDhzcPZaw5b-nrkX_Bp+OEYMg@mail.gmail.com></small>
 +
|-
 +
| [https://sourceforge.net/p/nmr-relax/mailman/message/36569049/ Aldino Viegas]
 +
| {{#time: Y-m-d H:i:s|Thu, 22 Sep 2011 10:32:26 +0200}}
 +
| <small><CAED9pY-=rnJ6EjMHAr-YDKOpf8JfG0unLP+JmZD-oOaGXL5Org@mail.gmail.com></small>
 +
|-
 +
| [https://sourceforge.net/p/nmr-relax/mailman/message/36569008/ Shantanu Bhattacharyya]
 +
| {{#time: Y-m-d H:i:s|Fri, 27 May 2011 10:16:48 +0200}}
 +
| <small><BANLkTim-PG-kZQGK7Fn+4SLXZ4vO-DxkYw@mail.gmail.com></small>
 +
|-
 +
| [https://sourceforge.net/p/nmr-relax/mailman/message/36568645/ Pierre-Yves Savard]
 +
| {{#time: Y-m-d H:i:s|Thu, 8 Jan 2009 17:41:20 +0100}}
 +
| <small><7f080ed10901080841p88137d1v8d04321393f12ab1@mail.gmail.com></small>
 +
|-
 +
| [https://sourceforge.net/p/nmr-relax/mailman/message/36568437/ Keith Constantine]
 +
| {{#time: Y-m-d H:i:s|Fri, 21 Mar 2008 21:17:33 +0100}}
 +
| <small><7f080ed10803211317k6fb586cfpc305eabbc80a5bf2@mail.gmail.com></small>
 +
|-
 +
| [https://sourceforge.net/p/nmr-relax/mailman/nmr-relax-users/thread/7f080ed10705071527r77c2dbe9oa180cd8e0bb8e69a%40domain.hid/#msg36568252 Clare-Louise Evans]
 +
| {{#time: Y-m-d H:i:s|Tue, 8 May 2007 00:27:59 +0200}}
 +
| <small><7f080ed10705071527r77c2dbe9oa180cd8e0bb8e69a@mail.gmail.com></small>
 +
|-
 +
| [https://sourceforge.net/p/nmr-relax/mailman/message/36570744/ Hongyan Li]
 +
| {{#time: Y-m-d H:i:s|Thu, 21 Dec 2006 19:53:51 +1100}}
 +
| <small><7f080ed10612210053l751e5fd8k96ee6097f0840849@mail.gmail.com></small>
 +
|-
 +
|}
  
a)  Find an initial diffusion tensor estimate (you can do this in
+
These will have lots of additional information.  This is just a selection of possibly the most useful messagesYou will soon see that this is a complicated topicNote that relax is capable of performing 100% of the functionality of:
relax by only using model m0).  This requires all non-mobile residues
 
and side chain spins to be excluded, and this can be problematicSee
 
the d'Auvergne and Gooley, 2008b paper at
 
http://dx.doi.org/10.1007/s10858-007-9213-3 for an example of the
 
catastrophic failure that this initial estimate can result inOr the
 
bacteriorhodopsin fragment of Korzhnev et al., 1999
 
(http://dx.doi.org/10.1023/a:1008356809071) where this complete
 
failure was earlier demonstrated.
 
  
b)  Optimise all of the model-free models from m0 to m9.  This
+
* [[Modelfree4]] (with or without the Fast-Modelfree GUI interface)
requires high precision optimisation, for a comparison of all the
+
* [[DASHA]]
softwares see the d'Auvergne and Gooley, 2008a model-free optimisation
+
* [[Tensor2]] and
paper at http://dx.doi.org/10.1007/s10858-007-9214-2.  Only relax and
+
* [[DYNAMICS]]
Dasha implement the full range of model-free models, though the models
 
m6, m7, and m8 cannot be used if only single field strength data is
 
used (m6 is the original 2-time scale motion model of Clore et al.,
 
1990).
 
  
c)  Eliminate failed models (this is only available in relax, see the
+
If you play with the optimisation settings you can even find identical results to within machine precision - relax can mimic these other softwares.
d'Auvergne and Gooley, 2006 model elimination paper at
 
http://dx.doi.org/10.1007/s10858-006-9007-z).
 
  
d)  Select the best model-free model for each spin system.  This again
+
=== Protocol ===
requires precision modern techniques, with the best being AIC model
 
select (see the d'Auvergne and Gooley, 2003 model-free model selection
 
paper at http://dx.doi.org/10.1023/A:1021902006114).  If you are
 
unaware that ANOVA statistics for model selection (hypothesis testing
 
via chi-squared, F- and t-tests) was abandoned by the field of model
 
selection over 100 years ago (a field which makes the NMR field look
 
very, very small), then you should really look at that paper.
 
  
e)  Optimise the global modelThis is the diffusion tensor plus the
+
The key is that the full analysis protocol is rather complicated - many people don't understand this - and that these softwares do not implement the full iterative protocolTherefore one either have to perform it manually or write a script to perform all of the steps.
model-free models for all spin systems.
 
  
f)  Check for convergence (identical chi-squared values to a previous
+
The protocol is described in the relax manual in figure 7.2 (http://www.nmr-relax.com/manual/diffusion_seeded_paradigm.html).
iteration, and not necessarily the last one).  If no, then go back to
 
b) and repeat.  Note that the chi-squared value can go up
 
significantly between iterations, but this is because the model is
 
simplifying itself at a much faster rate by loosing parameters - it's
 
Occam's razor at work.  Again see the d'Auvergne and Gooley, 2008b
 
paper at http://dx.doi.org/10.1007/s10858-007-9213-3 for figures
 
demonstrating this.  The concept as to what is happening during this
 
combined model-free optimisation and model selection algorithm is
 
described in the d'Auvergne and Gooley, 2007 MolBiosyst paper at
 
http://dx.doi.org/10.1039/b702202f. It can take up to 20 iterations
 
or more to reach convergence, depending upon the quality of the
 
relaxation data and the 3D structure or the system in study.
 
  
g)  Once steps a-f have been completed for all global models
+
In summary:
(characterised by the spheroid, prolate spheroid, oblate spheroid, and
 
ellipsoid diffusion tensors), then model selection between the
 
different global models needs to be performed.
 
  
h)  Monte Carlo simulations for error analysis must be performed at the end.
+
<ol style="list-style-type:lower-alpha">
 +
  <li>Find an initial diffusion tensor estimate (you can do this in relax by only using model m0).  This requires all non-mobile residues and side chain spins to be excluded, and this can be problematic.  See the [d'Auvergne and Gooley, 2008b] paper for an example of the catastrophic failure that this initial estimate can result in.  Or the bacteriorhodopsin fragment of [Orekhov et al., 1999] where this complete failure was earlier demonstrated.</li>
 +
  <li>Optimise all of the model-free models from m0 to m9.  This requires high precision optimisation, for a comparison of all the softwares see the [d'Auvergne and Gooley, 2008a] model-free optimisation paper.  Only relax and [[DASHA]] implement the full range of model-free models, though the models m6, m7, and m8 cannot be used if only single field strength data is used (m6 is the original 2-time scale motion model of [Clore et al., 1990]).</li>
 +
  <li>Eliminate failed models (this is only available in relax). See the [d'Auvergne and Gooley, 2006] model elimination paper.</li>
 +
  <li>Select the best model-free model for each spin system.  This again requires precision modern techniques, with the best being AIC model select (see the [d'Auvergne and Gooley, 2003] model-free model selection paper).  If you are unaware that ANOVA statistics for model selection (hypothesis testing via chi-squared, F- and t-tests) was abandoned by the field of model selection over 100 years ago (a field which makes the NMR field look very, very small), then you should really look at that paper.</li>
 +
  <li>Optimise the global model. This is the diffusion tensor plus the model-free models for all spin systems.</li>
 +
  <li>Check for convergence (identical chi-squared values to a previous iteration, and not necessarily the last one).  If no, then go back to b) and repeat.  Note that the chi-squared value can go up significantly between iterations, but this is because the model is simplifying itself at a much faster rate by loosing parameters - it's Occam's razor at work.  Again see the [d'Auvergne and Gooley, 2008b] paper for figures demonstrating this.  The concept as to what is happening during this combined model-free optimisation and model selection algorithm is described in the [d'Auvergne and Gooley, 2007] paper.  It can take up to 20 iterations or more to reach convergence, depending upon the quality of the relaxation data and the 3D structure or the system in study.</li>
 +
  <li>Once steps a-f have been completed for all global models (characterised by the spheroid, prolate spheroid, oblate spheroid, and ellipsoid diffusion tensors), then model selection between the different global models needs to be performed.</li>
 +
  <li>Monte Carlo simulations for error analysis must be performed at the end.</li>
 +
  <li>Elimination of failed Monte Carlo simulations is essential for keeping the errors to reasonable values for certain spin systems.  This is also a relax-only feature (see the [d'Auvergne and Gooley, 2007] model elimination paper).</li>
 +
</ol>
  
i)  Elimination of failed Monte Carlo simulations is essential for
+
These steps must be implemented independently of which software you use, as NONE implement the full protocol. Note however that the protocol I developed (in the [d'Auvergne and Gooley, 2007] theory paper and the [d'Auvergne and Gooley, 2008b] paper is fully implemented in relax, however this required multiple field strength data.
keeping the errors to reasonable values for certain spin systems.
 
This is also a relax-only feature (see the d'Auvergne and Gooley, 2007
 
model elimination paper at
 
http://dx.doi.org/10.1007/s10858-006-9007-z).
 
  
These steps must be implemented independently of which software you
+
This is a rather large script located at '''auto_anlayses/dauvergne_protocol.py'''.  This protocol is used by the GUI.  So one option would be to copy this '''auto_anlayses/dauvergne_protocol.py''' script and modify it for the figure 7.2 protocol.
use, as NONE implement the full protocol.  Note however that the
 
protocol I developed (in the d'Auvergne and Gooley, 2007 theory paper
 
at http://dx.doi.org/10.1039/b702202f and the d'Auvergne and Gooley,
 
2008b paper at http://dx.doi.org/10.1007/s10858-007-9213-3) is fully
 
implemented in relax, however this required multiple field strength
 
data.  This is a rather large script located at
 
auto_anlayses/dauvergne_protocol.py.  This protocol is used by the
 
GUI.  So one option would be to copy this
 
auto_anlayses/dauvergne_protocol.py script and modify it for the
 
figure 7.2 protocol.
 
  
 +
=== Warning ===
  
*** Note *** I must warn you about using single field strength data.
+
I must warn you about using single field strength data. It is now quite difficult to publish a model-free analysis with only single field strength data as most of the field know about the catastrophic analysis failures resulting in large amounts of artificial motion.  These failures can also be much more subtle.  Many reviewers will ask for such data to be collected as the results cannot not be trusted otherwise.  For a model-free analysis, it is almost
It is now quite difficult to publish a model-free analysis with only
+
essential to collect data at multiple field strengths, otherwise it can be sometimes impossible to distinguish between the anisotropic part of the Brownian tumbling of the molecule and internal motion - specifically due to the NH vectors in secondary structure elements all pointing in a similar direction.  I have a much better explanation, as well as citations to all the relevant literature in [d'Auvergne and Gooley, 2007].  In this paper, you will see reviewed both the artificial nanosecond motions of the [Schurr et al., 1994] paper and the artifical [[Rex|R<sub>ex</sub>]] motions of the [Tjandra et al., 1995] paper.
single field strength data as most of the field know about the
 
catastrophic analysis failures resulting in large amounts of
 
artificial motion.  These failures can also be much more subtle.  Many
 
reviewers will ask for such data to be collected as the results cannot
 
not be trusted otherwise.  For a model-free analysis, it is almost
 
essential to collect data at multiple field strengths, otherwise it
 
can be sometimes impossible to distinguish between the anisotropic
 
part of the Brownian tumbling of the molecule and internal motion -
 
specifically due to the NH vectors in secondary structure elements all
 
pointing in a similar direction.  I have a much better explanation, as
 
well as citations to all the relevant literature in:
 
  
d'Auvergne E. J., Gooley P. R. (2007). Set theory formulation of the
+
=== Recommendation ===
model-free problem and the diffusion seeded model-free paradigm. Mol.
 
Biosyst., 3(7), 483-494. (http://dx.doi.org/10.1039/b702202f)
 
  
In this paper, you will see reviewed both the artificial nanosecond
+
Finally, you will probably find it much easier to spend the 7-8 days collecting data at another field strength than to implement the [[#Protocol|protocol]] in a relax, [[Modelfree4]], or [[DASHA]] script (or via multiple iterations of the GUI programs), as well as study all of the relevant literature to understand all of the types of failures than only occurs with single field strength data.  With multiple field strength data you can perform {{relax developer link|username=semor|text=Sebastien Morin's}} consistency testing analysis in relax[Morin and Gagné, 2009] (see http://www.nmr-relax.com/manual/Consistency_testing.html).  That way you can see if your per-experiment temperature calibration and
motions of the Schurr 1994 paper and the artifical Rex motions of the
+
per-experiment temperature control techniques have works sufficiently well (http://www.nmr-relax.com/manual/Temperature_control_calibration.html) and if you have used long enough recycle delays.  Collecting data at a second field would probably save you significant amounts of time, and has the additional benefit that it would guarantee that the dynamics you see at the end will be real.  I cannot emphasize enough how important it is to collect data at multiple fields, most importantly the NOE and R<sub>2</sub> data.
Tjandra 1995 paper.
 
  
Finally, you will probably find it much easier to spend the 7-8 days
+
== References ==
collecting data at another field strength than to implement the
 
protocol of steps a-i in a relax, Modelfree4, or Dasha script (or via
 
multiple iterations of the GUI programs), as well as study all of the
 
relevant literature to understand all of the types of failures than
 
only occurs with single field strength data.  With multiple field
 
strength data you can perform Sebastien Morin's consistency testing
 
analysis in relax (http://dx.doi.org/10.1007/s10858-009-9381-4 and
 
http://www.nmr-relax.com/manual/Consistency_testing.html).  That way
 
you can see if your per-experiment temperature calibration and
 
per-experiment temperature control techniques have works sufficiently
 
well (http://www.nmr-relax.com/manual/Temperature_control_calibration.html)
 
and if you have used long enough recycle delays.  Collecting data at a
 
second field would probably save you significant amounts of time, and
 
has the additional benefit that it would guarantee that the dynamics
 
you see at the end will be real.  I cannot emphasize enough how
 
important it is to collect data at multiple fields, most importantly
 
the NOE and R2 data.
 
  
Regards,
+
*[*Clore et al., 1990] {{#lst:Citations|Clore90}}
 +
* [*d'Auvergne and Gooley, 2003] {{#lst:Citations|dAuvergneGooley03}}
 +
* [*d'Auvergne and Gooley, 2006] {{#lst:Citations|dAuvergneGooley06}}
 +
* [*d'Auvergne and Gooley, 2007] {{#lst:Citations|dAuvergneGooley07}}
 +
* [*d'Auvergne and Gooley, 2008a] {{#lst:Citations|dAuvergneGooley08a}}
 +
* [*d'Auvergne and Gooley, 2008b] {{#lst:Citations|dAuvergneGooley08b}}
 +
* [*Morin and Gagné, 2009] {{#lst:Citations|MorinGagné09}}
 +
* [*Orekhov et al., 1999] {{#lst:Citations|Orekhov99}}
 +
* [*Schurr et al., 1994] {{#lst:Citations|Schurr94}}
 +
* [*Tjandra et al., 1995] {{#lst:Citations|Tjandra95}}
 +
<HarvardReferences />
  
Edward
+
== See also ==
 +
[[Category:Model-free_analysis]]

Latest revision as of 08:46, 16 October 2020

Reference

https://sourceforge.net/p/nmr-relax/mailman/message/36569608/

The Question

Is possible to use relax to perform a model-free analysis using only relaxation data collected at a single magnetic field strength?

The answer is yes!

The subject of single field strength data has been discussed numerous times on this mailing list. It is recommended to read previous responses to questions relating to single field strength data, and look the other messages in those threads. You will find these discussions quite informative and highly detailed:

Person Message date Message ID
Danyun Zeng 2015-01-28 09:40:23 <CAED9pY-m5YEjN6KUX88s_yXLpY1BCudUaVvWWmC0U6WkteDnPA@mail.gmail.com>
Sean Leo Moro 2014-12-12 08:24:51 <CAED9pY9eAa7B+pC2+RU_pTXWR14fgtiMAiJdQ7rKw+aY6mALaA@mail.gmail.com>
Navratna Vajpai 2013-04-30 15:32:21 <CAED9pY-JcYpQtVpMe2dQf_iLvztQzVue=2d0mEQzYYnaYcRpnA@mail.gmail.com>
Shantanu Bhattacharyya 2013-01-07 10:29:28 <CAED9pY_aUzPH0OkZhTc2H=trV1VZ074XMcfT8SHYmKH5Bq6T3Q@mail.gmail.com>
Mengjun Xue 2012-09-06 12:14:19 <CAED9pY_R8s31My1w19uh5ss7Cjhh6WJLEs62sUUHsF8SxeEc7g@mail.gmail.com>
Fernando Amador 2012-05-11 07:34:27 <CAED9pY-gtfP__28VpkKgGgRVPAST0S0_Q+Hi94Zre0N=NfVLWg@mail.gmail.com>
Dhanasekaran Muthu 2012-01-31 14:45:22 <CAED9pY8UArGMiEf-3_fCjQrBJYPecu+BCGqkqtcSXwSix4e5cQ@mail.gmail.com>
Vitaly Vostrikov 2012-01-23 18:05:12 <CAED9pY8SGxe_rF6RKWmXJwyXjrDhzcPZaw5b-nrkX_Bp+OEYMg@mail.gmail.com>
Aldino Viegas 2011-09-22 08:32:26 <CAED9pY-=rnJ6EjMHAr-YDKOpf8JfG0unLP+JmZD-oOaGXL5Org@mail.gmail.com>
Shantanu Bhattacharyya 2011-05-27 08:16:48 <BANLkTim-PG-kZQGK7Fn+4SLXZ4vO-DxkYw@mail.gmail.com>
Pierre-Yves Savard 2009-01-08 16:41:20 <7f080ed10901080841p88137d1v8d04321393f12ab1@mail.gmail.com>
Keith Constantine 2008-03-21 20:17:33 <7f080ed10803211317k6fb586cfpc305eabbc80a5bf2@mail.gmail.com>
Clare-Louise Evans 2007-05-07 22:27:59 <7f080ed10705071527r77c2dbe9oa180cd8e0bb8e69a@mail.gmail.com>
Hongyan Li 2006-12-21 08:53:51 <7f080ed10612210053l751e5fd8k96ee6097f0840849@mail.gmail.com>

These will have lots of additional information. This is just a selection of possibly the most useful messages. You will soon see that this is a complicated topic. Note that relax is capable of performing 100% of the functionality of:

If you play with the optimisation settings you can even find identical results to within machine precision - relax can mimic these other softwares.

Protocol

The key is that the full analysis protocol is rather complicated - many people don't understand this - and that these softwares do not implement the full iterative protocol. Therefore one either have to perform it manually or write a script to perform all of the steps.

The protocol is described in the relax manual in figure 7.2 (http://www.nmr-relax.com/manual/diffusion_seeded_paradigm.html).

In summary:

  1. Find an initial diffusion tensor estimate (you can do this in relax by only using model m0). This requires all non-mobile residues and side chain spins to be excluded, and this can be problematic. See the [d'Auvergne and Gooley, 2008b] paper for an example of the catastrophic failure that this initial estimate can result in. Or the bacteriorhodopsin fragment of [Orekhov et al., 1999] where this complete failure was earlier demonstrated.
  2. Optimise all of the model-free models from m0 to m9. This requires high precision optimisation, for a comparison of all the softwares see the [d'Auvergne and Gooley, 2008a] model-free optimisation paper. Only relax and DASHA implement the full range of model-free models, though the models m6, m7, and m8 cannot be used if only single field strength data is used (m6 is the original 2-time scale motion model of [Clore et al., 1990]).
  3. Eliminate failed models (this is only available in relax). See the [d'Auvergne and Gooley, 2006] model elimination paper.
  4. Select the best model-free model for each spin system. This again requires precision modern techniques, with the best being AIC model select (see the [d'Auvergne and Gooley, 2003] model-free model selection paper). If you are unaware that ANOVA statistics for model selection (hypothesis testing via chi-squared, F- and t-tests) was abandoned by the field of model selection over 100 years ago (a field which makes the NMR field look very, very small), then you should really look at that paper.
  5. Optimise the global model. This is the diffusion tensor plus the model-free models for all spin systems.
  6. Check for convergence (identical chi-squared values to a previous iteration, and not necessarily the last one). If no, then go back to b) and repeat. Note that the chi-squared value can go up significantly between iterations, but this is because the model is simplifying itself at a much faster rate by loosing parameters - it's Occam's razor at work. Again see the [d'Auvergne and Gooley, 2008b] paper for figures demonstrating this. The concept as to what is happening during this combined model-free optimisation and model selection algorithm is described in the [d'Auvergne and Gooley, 2007] paper. It can take up to 20 iterations or more to reach convergence, depending upon the quality of the relaxation data and the 3D structure or the system in study.
  7. Once steps a-f have been completed for all global models (characterised by the spheroid, prolate spheroid, oblate spheroid, and ellipsoid diffusion tensors), then model selection between the different global models needs to be performed.
  8. Monte Carlo simulations for error analysis must be performed at the end.
  9. Elimination of failed Monte Carlo simulations is essential for keeping the errors to reasonable values for certain spin systems. This is also a relax-only feature (see the [d'Auvergne and Gooley, 2007] model elimination paper).

These steps must be implemented independently of which software you use, as NONE implement the full protocol. Note however that the protocol I developed (in the [d'Auvergne and Gooley, 2007] theory paper and the [d'Auvergne and Gooley, 2008b] paper is fully implemented in relax, however this required multiple field strength data.

This is a rather large script located at auto_anlayses/dauvergne_protocol.py. This protocol is used by the GUI. So one option would be to copy this auto_anlayses/dauvergne_protocol.py script and modify it for the figure 7.2 protocol.

Warning

I must warn you about using single field strength data. It is now quite difficult to publish a model-free analysis with only single field strength data as most of the field know about the catastrophic analysis failures resulting in large amounts of artificial motion. These failures can also be much more subtle. Many reviewers will ask for such data to be collected as the results cannot not be trusted otherwise. For a model-free analysis, it is almost essential to collect data at multiple field strengths, otherwise it can be sometimes impossible to distinguish between the anisotropic part of the Brownian tumbling of the molecule and internal motion - specifically due to the NH vectors in secondary structure elements all pointing in a similar direction. I have a much better explanation, as well as citations to all the relevant literature in [d'Auvergne and Gooley, 2007]. In this paper, you will see reviewed both the artificial nanosecond motions of the [Schurr et al., 1994] paper and the artifical Rex motions of the [Tjandra et al., 1995] paper.

Recommendation

Finally, you will probably find it much easier to spend the 7-8 days collecting data at another field strength than to implement the protocol in a relax, Modelfree4, or DASHA script (or via multiple iterations of the GUI programs), as well as study all of the relevant literature to understand all of the types of failures than only occurs with single field strength data. With multiple field strength data you can perform Sebastien Morin's consistency testing analysis in relax[Morin and Gagné, 2009] (see http://www.nmr-relax.com/manual/Consistency_testing.html). That way you can see if your per-experiment temperature calibration and per-experiment temperature control techniques have works sufficiently well (http://www.nmr-relax.com/manual/Temperature_control_calibration.html) and if you have used long enough recycle delays. Collecting data at a second field would probably save you significant amounts of time, and has the additional benefit that it would guarantee that the dynamics you see at the end will be real. I cannot emphasize enough how important it is to collect data at multiple fields, most importantly the NOE and R2 data.

References

  • [*Clore et al., 1990] Clore, G. M., Szabo, A., Bax, A., Kay, L. E., Driscoll, P. C., and Gronenborn, A. M. (1990). Deviations from the simple 2-parameter model-free approach to the interpretation of N-15 nuclear magnetic-relaxation of proteins. J. Am. Chem. Soc., 112(12), 4989-4991. (DOI: 10.1021/ja00168a070)
  • [*d'Auvergne and Gooley, 2003] d'Auvergne, E. J. and Gooley, P. R. (2003). The use of model selection in the model-free analysis of protein dynamics. J. Biomol. NMR, 25(1), 25-39. (DOI: 10.1023/a:1021902006114)
  • [*d'Auvergne and Gooley, 2006] d'Auvergne, E. J. and Gooley, P. R. (2006). Model-free model elimination: A new step in the model-free dynamic analysis of NMR relaxation data. J. Biomol. NMR, 35(2), 117-135. (DOI: 10.1007/s10858-006-9007-z)
  • [*d'Auvergne and Gooley, 2007] d'Auvergne, E. J. and Gooley, P. R. (2007). Set theory formulation of the model-free problem and the diffusion seeded model-free paradigm. Mol. BioSyst., 3(7), 483-494. (DOI: 10.1039/b702202f)
  • [*d'Auvergne and Gooley, 2008a] d'Auvergne, E. J. and Gooley, P. R. (2008). Optimisation of NMR dynamic models I. Minimisation algorithms and their performance within the model-free and Brownian rotational diffusion spaces. J. Biomol. NMR, 40(2), 107-119. (DOI: 10.1007/s10858-007-9214-2)
  • [*d'Auvergne and Gooley, 2008b] d'Auvergne, E. J. and Gooley, P. R. (2008). Optimisation of NMR dynamic models II. A new methodology for the dual optimisation of the model-free parameters and the Brownian rotational diffusion tensor. J. Biomol. NMR, 40(2), 121-133. (DOI: 10.1007/s10858-007-9213-3)
  • [*Morin and Gagné, 2009] Morin, S. and Gagné, S. (2009). Simple tests for the validation of multiple field spin relaxation data. J. Biomol. NMR, 45, 361-372. (DOI: 10.1007/s10858-009-9381-4)
  • [*Orekhov et al., 1999] Orekhov, V. Y., Korzhnev, D. M., Diercks, T., Kessler, H., and Arseniev, A. S. (1999). H-1-N-15 NMR dynamic study of an isolated alpha-helical peptide (1-36)bacteriorhodopsin reveals the equilibrium helix-coil transitions. J. Biomol. NMR, 14(4), 345-356. (DOI: 10.1023/a:1008356809071)
  • [*Schurr et al., 1994] Schurr, J. M., Babcock, H. P., and Fujimoto, B. S. (1994). A test of the model-free formulas. Effects of anisotropic rotational diffusion and dimerization. J. Magn. Reson. B, 105(3), 211-224. (DOI: 10.1006/jmrb.1994.1127)
  • [*Tjandra et al., 1995] Tjandra, N., Wingfield, P., Stahl, S., and Bax, A. (1996). Anisotropic rotational diffusion of perdeuterated HIV protease from N-15 NMR relaxation measurements at two magnetic. J. Biomol. NMR, 8(3), 273-284. (DOI: 10.1007/bf00410326)

<HarvardReferences />

See also