From: Joshua Adelman (jla65_at_pitt.edu)
Date: Tue Feb 15 2011 - 20:28:36 CST
Hi Flavio,
You could apply some sort of symmetry enforcing restraints (Jerome -- thanks for the correction/update about namd functionlity), but then you have to ask yourself how the restraints are impacting your results. Additionally, although I do not have extensive experience simulating models created via homology modeling, my personal opinion is that you need to be extremely cautious. Residues and loops that are incorrectly built might take a very long time to relax and might never sample the 'correct' conformation during the simulation.
The fact that some residues are in contact at an interface but equivalent residues at a different interface are not, could be caused by a number of different factors: the interfaces were poorly built (not through your fault, but perhaps because of the limits of homology modeling) and so you are seeing the beginning of a global instability in the tetramer; if you've only simulated a short amount of time, you are seeing a dynamics equilibrium between conformations and poor sampling keeps you from taking a realistic ensemble average of the behavior at the different interfaces; etc; etc.
You can certainly apply a restraint and then look at the 'mean' behavior, but you're going to have to tease out the effect of the restraint as well as convince yourself that you've sampled long enough to have confidence in the 'mean'.
Best of luck,
Josh
On Feb 15, 2011, at 6:20 PM, flavio seixas wrote:
> Hi Nikolaos and Josh,
>
> Well, I did explain a little part of the problem, maybe it had caused some kind of doubts about the purpose of my questions. Let me try to be more specific.
>
> Our group isolate an homotetramer enzyme and made biochemical characterization with it. But, this enzyme has a half life of few hours, regarding other enzymes from this family that have half life of months.
>
> Due to this short life, we were not able to crystallize the protein.
>
> The Clustaw alignment of the sequency of this enzyme with other enzymes from its family, shows that a lot of residues participating in contact between the monomers were lost by deletion or by replacement. We believe that this could be the reason of the protein instability, because it not works as dimers or monomers.
>
> I solved the structure by molecular modelling. The model is good as checked by procheck software.
>
> The proteins of this family are known to have identical monomers and identical contacts between them. We call this as non-crystallographic symmetry, as inplemented in Modeller, CNS, X-Plor and Refmac crystallographyc softwares.
>
> I made the modelling restraining only CA atoms in the subunits to be symetric.
>
> Regarding the model is good, the problem is that some residues are in contact at one interface, but not in others.
>
> My question is: What is the correct? The residue made the contact or not? Modelling by satisfaction of spatial restraining can not give me the answer.
>
> I am trying to solve this problem by MD.
>
> I expect that electrostatic interactions between residues of the interface atract them or repells them.
>
> I was wondering if applying any kind of "NCS" at the interface residues during MD simulation, could give me a "mean" scenario of the interactions, not a real 'non-crystallographic symmetry' model.
>
>
> Best regards,
>
> Flavio
>
>
> --- On Tue, 2/15/11, Nikolaos Glykos <glykos_at_mbg.duth.gr> wrote:
>
>> From: Nikolaos Glykos <glykos_at_mbg.duth.gr>
>> Subject: Re: namd-l: NCS
>> To: "flavio seixas" <oivalf_nix_at_yahoo.com>
>> Cc: "Joshua Adelman" <jla65_at_pitt.edu>, namd-l_at_ks.uiuc.edu
>> Date: Tuesday, February 15, 2011, 7:59 PM
>>
>> Hi Flavio,
>>
>> To add my twopence to what Josh said, even if you could
>> enforce (in the
>> form of restraints) the intramolecular symmetry, you would
>> be possibly
>> defeating the purpose of doing the MD in the first place :
>> the simulation
>> would simply reproduce your restraints and nothing more. To
>> give a far-fetched
>> example, if the homology-based model is totally wrong with
>> respect to the
>> relative orientation of the monomers about the
>> intramolecular symmetry axis,
>> by enforcing the (wrong) symmetry, you would be getting
>> back your original
>> mistakes. If the (unrestrained) dimer is not stable during
>> the MD, this
>> should be telling you something either about the quality of
>> the model, or the
>> assumed symmetry. Not restraining the symmetry has an added
>> advantage: you
>> can use convergence of the monomers' average structures to
>> gauge sufficient
>> sampling of your simulation.
>>
>> My twopence,
>> Nicholas
>>
>> ps. On a pedantic note, the usage of the term
>> 'non-crystallographic symmetry'
>> is possibly wrong both for the case of the homology-derived
>> model, and for
>> MD. It is (possibly) only relevant for the crystallographic
>> (template)
>> structure that you've used for homology modeling.
>>
>>
>>
>> On Tue, 15 Feb 2011 12:51:46 -0500, Joshua Adelman wrote:
>>> Hi Flavio,
>>>
>>> Could you expand on what you mean when you say you
>> want to enforce
>>> NCS on the dimer and explain a bit more what you
>> expect to see?
>>>
>>> In general though, in a MD simulation, each monomer is
>> going to
>>> fluctuate in some partially uncorrelated way with
>> regard to the other
>>> monomer, so physically you should not expect to see
>> perfect symmetry
>>> maintained throughout the simulation. The degree to
>> which the monomers
>>> are asymmetric in the simulation is dependent on a
>> complex set of
>>> conditions.
>>>
>>> Josh
>>>
>>>
>>> On Feb 15, 2011, at 6:04 AM, flavio seixas wrote:
>>>
>>>> Thanks for the suggestion.
>>>> Bit I think that kind of constraints will fix the
>> atoms at the interface, denying then to interact.
>>>> I need those atoms not fixed but interacting and
>> obeying a NCS constraint.
>>>>
>>>> flavio
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> One possibility is to use
>>>> constraints on
>>>> conskfile
>> interface.pdb
>>>> by defining a PDB file that contain amino acids at
>> the interface fix (e.g by adding 1 to temperature factor
>> col)
>>>> or
>>>> fixedAtomsFile interface.pdb
>>>>
>>>>
>>>> On Mon, Feb 14, 2011 at 2:09 PM, flavio seixas
>> <oivalf_nix_at_yahoo.com>
>> wrote:
>>>>
>>>> Hi,
>>>>
>>>>
>>>>
>>>> I have a doubt and I am interested in the opinion
>> of the more experts molecular dynamics people.
>>>>
>>>>
>>>>
>>>> I have already successful build a dimeric protein
>> by homology modeling using Non Crystallographic Symmetry
>> (NCS), i.e., the both monomers are identical. The A
>> aminoacid from subunity 1 contact the B aminoacid of
>> subunity 2 and vice-versa.
>>>>
>>>>
>>>>
>>>>
>>>> NCS is quite common in the proteins of this
>> family.
>>>>
>>>>
>>>>
>>>> After modeling step, I tried to minimize and
>> equilibrate the structure by NAMD.
>>>>
>>>>
>>>>
>>>> The output *.coord file shows different kinds of
>> contacts and they do not obey the NCS.
>>>>
>>>>
>>>>
>>>> My question is: Is there a way to run the
>> simulation making NAMD consider the NCS on atom position and
>> electrostatic interactions?
>>>>
>>>>
>>>>
>>>> If the answer is yes, How do I do it on the
>> script????
>>>>
>>>>
>>>>
>>>> Thanks for any help.
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> Flavio
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> Dr. Flavio Augusto Vicente Seixas
>>>>
>>>> Professor Adjunto do Departamento de Bioquímica -
>> DBQ
>>>>
>>>> Universidade Estadual de Maringá - UEM
>>>>
>>>> Av. Ângelo Moreira da Fonseca, 1800 - Zona 7
>>>>
>>>> 87506-370 - UMUARAMA - PR - BRAZIL
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>
>>
>
>
>
>
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:56:39 CST