Well, yes, it is a real problem.
We sometimes do not know. If you consider the simplicity
of the force field, then even with all the refinement
efforts we cannot have a perfect force field.
We have discussed the issues little more in these
papers:
The DNA and RNA sugar-phosphate backbone emerges as the key player. An
overview of quantum-chemical, structural biology and simulation studies
Author(s): Sponer, Jiri; Mladek, Arnost; Sponer, Judit E.; et al.
Source: PHYSICAL CHEMISTRY CHEMICAL PHYSICS Volume: 14 Issue: 44
Pages: 15257-15277 DOI: 10.1039/c2cp41987d Published: 2012
Molecular dynamics simulations of G-DNA and perspectives on the simulation
of nucleic acid structures
Author(s): Sponer, Jiri; Cang, Xiaohui; Cheatham, Thomas E., III
Source: METHODS Volume: 57 Issue: 1 Pages: 25-39 DOI:
10.1016/j.ymeth.2012.04.005 Published: MAY 2012
1) I would use a long simulation (as long as possible, and multiple)
of the native structure
as a kind of benchmark. If you see evident deviations
from the X-ray structure accumulating over the time,
then most likely the simulation is entering a phase
of force field problems.
Of course, there could be errors (and sometimes
considerable such as syn/anti bias) also in the exp.
structure and there could be also crystal packing
issues. Backbone is barely visible at 2.3 A.
So the simulation can also respond to experimental errors!!
(we have documented such cases in the past).
We are often inspecting electron densities, checking
packing, making bioinformatics analyses.
(I would be happy to know in case you see some big
problems, it could help us to better understand the
limits.)
2) Then I would make the mutated simulations.
I would assume the native and mutated systems
are affected by similar errors, so the
difference in behaviour between mutated and native simulations
could indeed by the real effect. Of course, there is no
100% guarantee of that, but it is a very plausible
interpretation. The force field description of
base pairing and stacking is quite good so the
simulation should correctly respond to base substitutions.
Best wishes, Jiri
On Sat, 30 Mar 2013, Asmita Gupta wrote:
> Date: Sat, 30 Mar 2013 23:57:27 +0530
> From: Asmita Gupta <asmita4des.gmail.com>
> Reply-To: AMBER Mailing List <amber.ambermd.org>
> To: AMBER Mailing List <amber.ambermd.org>
> Subject: Re: [AMBER] conformational sampling in non helical RNA structures
>
> Thanks a lot for the response...i read the article you mentioned.
>
> I have used Amber12 force field ff12SB (with OL3 corrections) for all my
> simulations (effect of different force fields on RNA systems has been
> documented in your recent articles).
>
> As i mentioned that my systems are mutants of the original crystal
> structure, how can we suggest whether any conformational difference
> observed is a result of force field limitations or is an actual consequence
> of mutations. (Though, all recent papers recommend ff12SB with OL3)
>
> I am a bit confused in this regard....
>
> Thanks
>
> Asmita
>
>
> On Sat, Mar 30, 2013 at 9:41 PM, Jiri Sponer <sponer.ncbr.muni.cz> wrote:
>
>> This is not easy question.
>> 2.3 A could be more or less good start structure.
>> Modifications make a sense, MD should capture their
>> effect.
>> It depends if they are isosteric or not.
>> However, if and when to see changes, it is case by case.
>> With increasing the simulation time, the probability
>> to see something increases, but also to see force field
>> problems.....
>>
>> I would tell multiple 100-500 ns simulations are good
>> option.
>>
>> Look at this papar where we tried to see similar issues,
>> expanding in some cases above microseconds.
>> Molecular Mechanism of preQ(1) Riboswitch Action: A Molecular Dynamics
>> Study
>> Author(s): Banas, Pavel; Sklenovsky, Petr; Wedekind, Joseph E.; et al.
>> Source: JOURNAL OF PHYSICAL CHEMISTRY B Volume: 116 Issue: 42 Pages:
>> 12721-12734 DOI: 10.1021/jp309230v Published: OCT 25 2012
>>
>> Note that RMSd is a poor descriptor, it tells nothing, you must analyze
>> the details of the structure.
>>
>>
>> Best wishes, Jiri
>>
>>
>>
>> On Sat, 30 Mar 2013, Asmita Gupta wrote:
>>
>>> Date: Sat, 30 Mar 2013 21:14:39 +0530
>>> From: Asmita Gupta <asmita4des.gmail.com>
>>> Reply-To: AMBER Mailing List <amber.ambermd.org>
>>> To: AMBER Mailing List <amber.ambermd.org>
>>> Subject: [AMBER] conformational sampling in non helical RNA structures
>>>
>>> Dear Users,
>>>
>>> I am running an MD simulation on an RNA aptamer, (2.3 Ang resolution, 20
>>> residues). Apart from the native state, i have generated 3 mutant forms
>> of
>>> the original crystal structure. These mutations correspond to either
>> single
>>> base mutations or base pair substitutions.
>>>
>>> As i am not sure about the behaviour of these mutant forms, current
>>> simulation time is limited to 100ns. I don't know whether extending the
>>> simulation time will bring the system to convergence or the system will
>>> fall apart.
>>>
>>> My purpose is to generate a set of ensembles for analysis of
>>> conformational differences in mutant forms. Can you please suggest what
>>> should be done to enhance sampling. Is extending the time scale is a
>> viable
>>> solution(in case of above uncertainty)
>>>
>>> I calculated RMSD for all heavy atoms. The curve was stable after 70ns,
>>> but RMSD for the loops seemed to increase after 90ns.
>>>
>>> Thanks
>>>
>>> Asmita
>>> _______________________________________________
>>> AMBER mailing list
>>> AMBER.ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Mar 30 2013 - 12:30:02 PDT
