[gmx-users] g_rotacf

Xavier Periole X.Periole at rug.nl
Fri Sep 5 10:08:22 CEST 2008


On Thu, 4 Sep 2008 21:13:39 -0400
  "rams rams" <rams.crux at gmail.com> wrote:
> Hi Xavier,
> 
> I have a question about the accuracy of the correlation times obtained using
> gromacs tools. (I am getting these by integrating the output .xvg file of
> g_rotacf using g_analyze). Since the experimental values are not available
> to compare, I am not so sure about the correlation times I am getting. They
> are of the order of 0.4 to 0.8ns for various N-H vectors. I believe the
> values depends upon the range of time frames we use to fit with the legendre
> polynomials. I am using second order legendre polynomial to fit the frames.
> Can you give me any general suggestions to follow to obtain the reasonable
> correlation time values with MD simulations.
The range of values depend on the molecule you look at and the solvent
viscosity. A protein with 50-100 AA would have an experimental rotational
correlation time around 3-5 ns. In MD you find in general faster motion (1-3
ns may be) which is due to the solvent and may be solvent-protein interactions
that are a bit too weak, but this is not completely understood.
I am more concerned about the procedure you actually use to obtain your value.
The Legendre polynomial should be of the order 2 if you compare to NMR
relaxation data, this is ok, but you do not use to fit the data! You use it
to build the correlation function. Then you fit the correlation function
with a function that would describe the motion you look at! For an NH vector
of the backbone there are several motion model that one has to decide
which one is best. Have a look at that paper. I did not read it but it
talks about model to fit NMR data and should contain a lot of reference
for you to begin with. DOI: 10.1021/jp8038576.

Then you should have look at your correlation functions. Are they converged
(reached a plateau)? How do they look?

You can also use the g_dipole to get the auto correlation function of the
molecule dipole ...
> Ram.
> 
> On Thu, Sep 4, 2008 at 11:02 AM, Xavier Periole <X.Periole at rug.nl> wrote:
> 
>> On Thu, 4 Sep 2008 09:45:34 -0400
>>  "rams rams" <rams.crux at gmail.com> wrote:
>>
>>> Hi Xavier,
>>>
>>> I am extremely sorry for incomplete information. But this is a follow up
>>> to
>>> my previous two emails. Unfortunately I havent received any suggestions
>>> for
>>> them and it appears like you too missed them. For more clarity I am
>>> pasting
>>> them here again:
>>>
>> Well if you do not get an answer that is probably that your post is not
>> really asking a quick "answerable" question and you might want to
>> re-formulate your post or dig a bit to make things easier.
>>
>>> To calculate the rotational auto correlation functions, the command
>>> mentioned in the maual is:
>>>
>>> g_rotacf -P 1 -nparm 2 -fft -n index -o .xvg -fa  -beginfit -endfit
>>>
>>> what are the -nparm and -fa options are meant for ?
>>>
>> This is an old exemple ... a bad one! Those options do not seem to exist
>> anymore! Never knew they existed!
>> you can grep "\-nparm" and "\-fa" in tools/*.c and you'll see they
>> only appear in gmx_rotacf.c
>> anyways These things are probably taken care of more invisibly.
>> -nparm seems to give the number of parameters in the function to fit
>> -fa the function to fit
>>
>> now: -nparm do not exist and -fa is -fitfn, most likely!
>>
>> This exemple should be removed and replaced! Just look at the
>> options\listed when you type g_rotacf -h.
>>
>>> Also do we need to use -fitfn option to obtain the rotational auto
>>> correlation functions ??
>>>
>> No, -fitfn defines the type of function you'd use to fit the correlation
>> function, which itself is written cause -o rotacf.xvg ...
>>
>> XAvier.
>>
>>
>>> Ram.
>>>
>>>
>>>
>>>
>>>
>>> Dear users,
>>>
>>> I have given a command like the following to calculate the rotational auto
>>> correlation function:
>>>
>>> g_rotacf -f .trr -s .tpr -P 2 -fft -o .xvg -b 0000 -e 10000  -n .ndx -d
>>>
>>> I want to use the second order Legendre polynomial to fit. I integrated
>>> the
>>> resulting .xvg file, to obtain the correlation time using:
>>>
>>> g_analyze -f .xvg  -integrate
>>>
>>> The output is the following. Could some one help me in understanding it
>>> betterly.
>>>
>>> Calculating the integral using the trapezium rule
>>> Integral 1   120.92840  +/-    0.00000
>>>                                     std. dev.    relative deviation of
>>>                      standard       ---------   cumulants from those of
>>> set      average       deviation      sqrt(n-1)   a Gaussian distribition
>>>                                                     cum. 3   cum. 4
>>> SS1   2.428352e-02   2.972849e-02   4.204243e-04       9.301   120.228
>>>
>>>
>>> Which value corresponds to correlation time and in which units (its in ps
>>> I
>>> suppose) ?
>>>
>>> Thanks in advance.
>>>
>>> Ram.
>>>
>>>
>>> On Thu, Sep 4, 2008 at 2:32 AM, Xavier Periole <X.Periole at rug.nl> wrote:
>>>
>>>  On Wed, 3 Sep 2008 23:40:14 -0400
>>>>  "rams rams" <rams.crux at gmail.com> wrote:
>>>>
>>>>  Hi,
>>>>>
>>>>> I am so surprised for not finding any one who have better experience
>>>>> with
>>>>> g_rotacf. I have been playing around with it and the time correlation
>>>>> value
>>>>> I got by g_rotacf is so small in comparison to the time correlation
>>>>> value
>>>>> I
>>>>> calcualted using the hydrodynamic radius of the protein. The value is
>>>>> nearly
>>>>> 10 times less. Can some one give me a better idea about g_rotacf.
>>>>>
>>>>>  Many people have certainly used g_rotacf to get ACFs of different
>>>> observables.
>>>>
>>>> Anyways the way you describe your system, command line and your problem
>>>> does
>>>> does not help anyone to help you. Read your message above and think about
>>>> what
>>>> you would answer! You've played around with g_rotacf therefore you know
>>>> it
>>>> is not straightforward to give you the magic command.
>>>>
>>>> XAvier.
>>>>
>>>>
>>>>> Ram.
>>>>>
>>>>>
>>>> -----------------------------------------------------
>>>> XAvier Periole - PhD
>>>>
>>>> Molecular Dynamics Group / NMR and Computation
>>>> University of Groningen
>>>> The Netherlands
>>>> -----------------------------------------------------
>>>> _______________________________________________
>>>> gmx-users mailing list    gmx-users at gromacs.org
>>>> http://www.gromacs.org/mailman/listinfo/gmx-users
>>>> Please search the archive at http://www.gromacs.org/search before
>>>> posting!
>>>> Please don't post (un)subscribe requests to the list. Use the www
>>>> interface
>>>> or send it to gmx-users-request at gromacs.org.
>>>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>>>>
>>>>
>> -----------------------------------------------------
>> XAvier Periole - PhD
>>
>> Molecular Dynamics Group / NMR and Computation
>> University of Groningen
>> The Netherlands
>> -----------------------------------------------------
>> _______________________________________________
>> gmx-users mailing list    gmx-users at gromacs.org
>> http://www.gromacs.org/mailman/listinfo/gmx-users
>> Please search the archive at http://www.gromacs.org/search before posting!
>> Please don't post (un)subscribe requests to the list. Use the www interface
>> or send it to gmx-users-request at gromacs.org.
>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>>

-----------------------------------------------------
XAvier Periole - PhD

Molecular Dynamics Group / NMR and Computation
University of Groningen
The Netherlands
-----------------------------------------------------



More information about the gromacs.org_gmx-users mailing list