[gmx-users] Validation of molecular dynamic simulation results

Tue Feb 11 17:43:07 CET 2014

On 2/11/14, 6:43 AM, ananyachatterjee wrote:
> Dear Sir,
>
> My protein contains two domain and upon GTP hydrolysis, the protein under goes
> conformational change and it shows inter-domain movement. I did few biochemical
> test which showed me that the upon a mutation in the conserved residue (which is
> localed in the chain connecting the two domain)the GTPase activity is reduced
> and also it has effected the cell growth. Now to see whether the mutation is
> some how effecting the inter domain movement of the protein(which is a important
> property of the protein for it function) I performed the simulations keeping the
> parameters close to the biochemical experiments(like con. of ions, temp. etc) .
> Now when I presented my work in a conference people asked me about its
> authenticity, how much close it is to the reality and whether it is
> reproducible. I want to know how to verify my simulations.
>

Reproducibility is an important issue, especially in simulations.  You should 
read the following page:

http://www.gromacs.org/Documentation/Terminology/Reproducibility

Related issues are convergence and sampling.  If I had to guess, I would assume 
that's what the main criticism from the conference is.  A single simulation may 
not be at all representative of reality; in fact, it may be an outlier of the 
real behavior.  A simulation that is long enough such that the time average of 
the trajectory reflects the ensemble average is ergodic.  That's hard to do and 
hard to prove, so typically multiple simulations are conducted and analyzed.  If 
they all converge towards the same behavior, the outcome is much more reliable.

Beyond that, there is no program that one can magically run to ensure that a 
simulation is valid or reliable.  Correspondence with experimental observations 
is critical to making such arguments, though you haven't provided much concrete 
detail there in terms of what you can look at.

-Justin

> Thanks in advance
>
> On Tue, 11 Feb 2014 11:24:53 +0100, João Henriques wrote:
>> Dear Ananya,
>>
>> Sorry but I don't understand what you're saying. What do you mean by
>> "molecular
>> movement of my protein"? Do you mean diffusion? Please be more specific.
>> You also mention "authenticity", but I don't think that's what you meant...
>>
>> You have done PCA, RMSD and RMSF analyses. That's nice and all, but do you
>> have a reason for doing them? What I meant with my previous email is
>> that *there
>> is no fixed recipe that one must always follow in order to validate their
>> simulations*. The analyses your simulation requires in order to be
>> validated depend on the simulation purpose and what experimental data you
>> have to compare them to.
>>
>> Take this overly simplistic example: Imagine I've performed a simple
>> "protein in water" MD simulation. The protein is well behaved (stable
>> native structure) and there is a high resolution experimental structure of
>> it. Radius of gyration, RMSD, RMSF and DSSP analyses would be good
>> candidates to show whether my force field, settings and parameters are
>> adequate or not in maintaining the well known native 3D structure.
>>
>> Like I said before, you're the one that must be familiar with your system,
>> simulation and overall project aim. I am happy to help you in identifying
>> the tool you need to perform a certain analysis/study or even helping
>> understanding any error or problem with it. However I cannot do your own
>> work in identifying what needs to be done and where you're heading.
>>
>> Best regards,
>> João
>>
>>
>> On Tue, Feb 11, 2014 at 9:46 AM, ananyachatterjee <
>> ananyachatterjee at iiserkol.ac.in> wrote:
>>
>>> On Tue, 11 Feb 2014 09:05:19 +0100, João Henriques wrote:
>>>
>>>> Dear Ananya,
>>>>
>>>> Shouldn't this be something you already had in mind even before attempting
>>>> to simulate? Usually, a simulation is a means to an end. What is your end,
>>>> ie. what made you do this simulation? The motivation behind your
>>>> simulation
>>>> is usually what will determine what type of validation it requires. Sure,
>>>> there are quasi-standard analysis and sanity checks that one almost always
>>>> uses to ensure the simulation is relevant, but you can find those in any
>>>> given MD tutorial available online. For this, Google is your friend and
>>>> you
>>>> also have quite popular tutorials such as Justin's for example:
>>>>
>>>>
>>>>
>>>> http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/
>>>>
>>>> I suggest you sit down and think a bit about what are you attempting to
>>>> achieve with your simulation. Pinpoint what type of analysis you need and
>>>> then take a careful look at the Gromacs manual. There you will find all
>>>> you
>>>> need to know about the analysis tools available to you. They're not
>>>> guaranteed to cover all your needs, but they usually more than suffice in
>>>> providing what most researchers need to validate their simulations.
>>>>
>>>> Best regards,
>>>> João
>>>>
>>>>
>>>> On Tue, Feb 11, 2014 at 8:02 AM, ananyachatterjee <
>>>> ananyachatterjee at iiserkol.ac.in> wrote:
>>>>
>>>>  Hello everyone,
>>>>>
>>>>> I have done a set of molecular dynamic simulation of my protein and its
>>>>> mutated structure, now please tell me how should I validate the
>>>>> simulation
>>>>> results or structures. Whether I can reproduce the same simulations.
>>>>>
>>>>> Thank you in advance
>>>>>
>>>>> --
>>>>> Ananya Chatterjee,
>>>>> Senior Research Fellow (SRF),
>>>>> Department of biological Science,
>>>>> IISER-Kolkata.
>>>>> --
>>>>> Gromacs Users mailing list
>>>>>
>>>>> * Please search the archive at http://www.gromacs.org/
>>>>> Support/Mailing_Lists/GMX-Users_List before posting!
>>>>>
>>>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>>>
>>>>> * For (un)subscribe requests visit
>>>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>>>>> send a mail to gmx-users-request at gromacs.org.
>>>>>
>>>>>
>>> Dear João,
>>>
>>> Thank you for your reply, I was intended to seen the difference in
>>> molecular movement of my protein upon mutation. My protein is a GTPase
>>> protein and I have done simulation in presence of GTP, GDP and GDP+Pi for
>>> both the wild type and the mutated protein. Then I have done principle
>>> component analysis and RMSD & RMSF comparison of the trajectories. Now I
>>> want to check the authenticity of the set of simulation and whether they
>>> can be reproduce to validate my results.
>>>
>>> Please kindly help me in this regard.
>>>
>>> --
>>> Ananya Chatterjee,
>>> Senior Research Fellow (SRF),
>>> Department of biological Science,
>>> IISER-Kolkata.
>>> --
>>> Gromacs Users mailing list
>>>
>>> * Please search the archive at http://www.gromacs.org/
>>> Support/Mailing_Lists/GMX-Users_List before posting!
>>>
>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>
>>> * For (un)subscribe requests visit
>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>>> send a mail to gmx-users-request at gromacs.org.
>>>
>

-- 
==================================================

Justin A. Lemkul, Ph.D.
Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================