AMBER Archive (2009)

Subject: Re: [AMBER] Creating conditions for biased MD

From: Francesco Pietra (chiendarret_at_gmail.com)
Date: Wed Apr 22 2009 - 05:16:11 CDT

I have insisted with the impose command as the manual does not rule
out reading from a pdb. It worked nicely to give a nearly perfect
alpha helix from the isolated bent helix. It worked provided the atom
names are enclosed within quotation marks (as in my previous mail and
as Scott Brozell suggested long ago). Without quotation marks the
bending increased from the original 40 degrees to ca 90 degrees. The
leap.log dis not give any warning about impose in either case.

The impose command - issued as above on the complex of helices
(separated from each other, as each one is capped at both extremes)
did not work: the bent helix remained unaltered, while another helix
was removed from the complex of helices and placed at right angle with
respect to the main axis of the complex.

I have not investigated further what happened. What I should be able
to do now is renumbering the straightened isolated helix (leap has
renumbered it starting from 1) so that it fits into the complex at
about its original place. Big steric crashes are expected. Do you
think that, after renumbering as said above, the leap command

model3 = combine { model1 model2 }

could be used to avoid clashes?

If clashes will not be avoided (or if leap is unable to adjust the
system even starting from a situation of clashes), I am prepared to
move to suggestion two: restraints.

thanks

francesco

On Tue, Apr 21, 2009 at 9:44 PM, Carlos Simmerling
<carlos.simmerling_at_gmail.com> wrote:
> I was thinking you could do that if it was only the helix, and I meant
> to build a perfect helix from sequence. not sure if impose works when
> reading a pdb. I meant to build the helix alone and use that for the
> reference, but if the system is more complicated then you should
> probably use restraints.
>
>
> On Tue, Apr 21, 2009 at 2:40 PM, Francesco Pietra <chiendarret_at_gmail.com> wrote:
>> On Mon, Apr 20, 2009 at 8:46 PM, Carlos Simmerling
>> <carlos.simmerling_at_gmail.com> wrote:
>>> I think it might be easier with dihedral angle restraints. it depends
>>> on the rest of the system- if it's just the one helix, you could use
>>> the impose command in leap to make a continuous helix as a reference
>>> for SMD.
>>
>> I begun from the last suggestion, recreating the protein (made of
>> capped helices which had been before isolated from the whole protein),
>> with commands "model1  = loadpdb" "model2 = load pdb" (where the two
>> models had been aligned before)  and "model3 = combine" the protein
>> was inserted into a lipidic membrane.
>>
>> Then command
>>
>> impose model3 { 225 226 227 228 229 230 }  {  {  "N" "CA" "C" "N"
>> -40.0 }  {  "C" "N" "CA" "C" -60.0 } }
>>
>> Then, with command "solvatebox" it was hydrated. The prmtop and inpcrd
>> opened in a viewer, WAT and lipid removed, the bent helix looks like
>> (grossly) unaffected. The bending angle is still ca 40 degrees.
>>
>> Should straightening of the helix have been expected at this stage or
>> the impose command will be felt on running MD simulations? Or, have i
>> misunderstood everything?
>>
>> thanks for having a look at this message
>>
>> francesco
>>
>>
>>>
>>>
>>>
>>> On Mon, Apr 20, 2009 at 1:31 PM, Francesco Pietra <chiendarret_at_gmail.com> wrote:
>>>> Carlos: late in thanking you. I had to fix hardware and software problems.
>>>>
>>>> Complex:
>>>>
>>>> phi Gly227C  Gly228N  Gly228CA Gly228C  ca 78 degrees (positive)
>>>>
>>>> psi Gly228N  Gly228CA  Gly228C  Gln229N  -19 degrees
>>>>
>>>> In a cartoon view, two cylinders make an angle of ca 40 degrees,
>>>> interconnected by 227 228. I would like to approximately straighten
>>>> that, to get a continuous cylinder. In a helix view, phi should be
>>>> turned anticlockwise to reach at least the value -80 degrees (the
>>>> value of corresponding nearby dihedral is just -80 degrees, while
>>>> farther away it takes normal values, -57 degrees). psi, at least, has
>>>> the correct sign.
>>>>
>>>> The bent helix is capped on both sides, but there are other similar,
>>>> capped helices around, so that I guess the work should be carried out
>>>> on the whole model. If I could restrain the capping groups alone of
>>>> the bent helix (while finding a way that helicity, where correct, is
>>>> conserved) I could imagine that straightening could occur. Otherwise,
>>>> if all amino acids, except 227 228 of the bent helix, are restrained,
>>>> how could the helix get straightened? Is that a task for SMD at all?
>>>>
>>>> thanks
>>>> francesco
>>>>
>>>>
>>>>
>>>>
>>>> On Mon, Mar 30, 2009 at 1:56 PM, Carlos Simmerling
>>>> <carlos.simmerling_at_gmail.com> wrote:
>>>>> francesco, I think dihedral restraints may be the only way to go without a
>>>>> helical ref structure. you'd definitely want to also restrain all other
>>>>> residues in this helix to maintain it, or else you might just move the bend.
>>>>> when changing the restraints over time, give thought to which direction to
>>>>> rotate, which depends on the initial conformation of the Gly residues, if
>>>>> they are currently adopting positive phi values then it's more complex than
>>>>> just going from pp2 to alpha, for example, where you can just reduce psi.
>>>>>
>>>>> On Mon, Mar 30, 2009 at 6:16 AM, Francesco Pietra <chiendarret_at_gmail.com>wrote:
>>>>>
>>>>>> Hi:
>>>>>> I would like to modify the conformation of a protein at one helix,
>>>>>> which is bent at a region of three amino acids, Ile, Gly, Gly. Viewed
>>>>>> in cartoon representation, it is constituted of two straight portions
>>>>>> interconnected through a largely non helical set of the three amino
>>>>>> acids.
>>>>>>
>>>>>> I would appreciate suggestions how to get the three aa taking part to
>>>>>> the well ordered helical conformation, so that what is now (in
>>>>>> cartoon) two straight portions interconnected by something like a loop
>>>>>> becomes a wholly straitened motif. Rotation about dihedrals? Make the
>>>>>> process with the isolated helix or in the whole context of the
>>>>>> protein?
>>>>>>
>>>>>> I guess that steered MD (on which I have no experience) is the
>>>>>> approach in Amber, though I wonder how to provide the target.
>>>>>> Possibly, any biased MD should be carried out in explicit medium. In
>>>>>> my hands, continuum models were unsuccessful with this protein.
>>>>>>
>>>>>> If these are not such naive questions to merit attention, thanks
>>>>>>
>>>>>> francesco pietra
>>>>>>
>>>>>> _______________________________________________
>>>>>> AMBER mailing list
>>>>>> AMBER_at_ambermd.org
>>>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>>>>
>>>>> _______________________________________________
>>>>> AMBER mailing list
>>>>> AMBER_at_ambermd.org
>>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>>>
>>>>
>>>> _______________________________________________
>>>> AMBER mailing list
>>>> AMBER_at_ambermd.org
>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>>
>>>
>>> _______________________________________________
>>> AMBER mailing list
>>> AMBER_at_ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER_at_ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
>
> _______________________________________________
> AMBER mailing list
> AMBER_at_ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber