Hi Pierpaolo,
This is good to hear. I took a closer look at this over the weekend and I
agree with Scott that this looks to be kosha since this is an internal
'restraint' and so the X, Y and Z components are simply additive. The RMS
fluctuations look pretty big though - can you post Etot vs time from your
output file please (as well as your mdin).
We could implement this in a general fashion in the code fairly easily if
there is a defined need for it. Note the confusion arose because you are
continually referring to a 'constraint' here which, if I understand
correctly what you are doing this is not. This is a restraint, with a
corresponding restraint energy and restraint force. This is VERY different
to a constraint and the underlying math is very different. I would take
the time to read up on the differences between restraints and constraints
before you try to use this for calculating things like permeabilities. If
you are just using it to get you a reasonable starting structure then this
is fine.
All the best
Ross
On 6/23/14, 3:20 AM, "Pierpaolo Cacciotto"
<pierpaolo.cacciotto.dsf.unica.it> wrote:
>Dear Ross and Scott,
>
>thank you very much for your comments and suggestions.
>I run a 1ns nve simulation during this weekend as you suggested and the
>energy is conserved. I hope the run was not too short for such a test.
>
>I must say that I am using this z-constraint because when I inserted a
>protein in a POPE membrane there were gaps between the protein and the
>lipids.
>I just wanted the membrane to get close to the protein before adding
>water, counter-ions, etc. I didn't mean to run a full simulation in these
>conditions.
>But maybe it would be nice to have an option to set some constraints
>direction (I knew my change was not too elegant...).
>
>Here the result of the nve test.
>
>--------------------------------------------------
>
> R M S F L U C T U A T I O N S
>
>Etot = 20.2981 EKtot = 282.4004 EPtot =
>284.5927
>
>---------------------------------------------------
>
>
>Thank you again for your comments.
>
>Regards,
>Pierpaolo
>
>
>> Of course, because I'm not right in the head, I would prefer defining a
>> local orthonormal coordinate system myself so one could do this
>> arbitrarily
>> rather than strictly along the x, y, and z axes...
>>
>>
>> On Fri, Jun 20, 2014 at 9:13 AM, Scott Le Grand <varelse2005.gmail.com>
>> wrote:
>>
>>> Looking at the rest of the surrounding code, I *think* this works
>>> because
>>> the three axes are independent and additive components (if this were
>>> otherwise, Ross is dead-on about breaking detailed balance). They just
>>> happen to have the same weight according to the rules we specify.
>>>
>>> That said, of course validate it with some NVE runs...
>>>
>>> And then if this were actually something people needed, I would
>>>formally
>>> add a constraint type with per axis components rather than kludge the
>>> code
>>> like this...
>>>
>>>
>>>
>>>
>>> On Fri, Jun 20, 2014 at 8:57 AM, Ross Walker <ross.rosswalker.co.uk>
>>> wrote:
>>>
>>>> Hi Pierpaolo
>>>>
>>>> ---
>>>> I modified the file constraints.F90 in
>>>>/usr/local/amber14/src/pmemd/src
>>>> as
>>>> follows
>>>>
>>>> wt = weit(j) # line 245
>>>> ax = x(1, i) - xc(1, i)
>>>> ay = x(2, i) - xc(2, i)
>>>> az = x(3, i) - xc(3, i)
>>>> wx = 0.d0 # here I changed the values of wx and wy
>>>> wy = 0.d0 #
>>>> wz = wt * az
>>>>
>>>> ---
>>>>
>>>> Have you worked through the math for this to make sure it is valid and
>>>> obeys detailed balance? I haven't had a chance to myself but I cannot
>>>> believe that one can simply zero the constraint here and not do
>>>> anything
>>>> with regards to balancing the gradients etc. If you run your
>>>>simulation
>>>> with NVE does it conserve energy with this modification? If not then
>>>> you
>>>> can't simply make such a change. Even if it appears to run stably all
>>>> of
>>>> your statistical distributions will come out incorrect.
>>>>
>>>> You need to work through the math first and verify this is correct
>>>> before
>>>> simply making edits in the code.
>>>>
>>>> All the best
>>>> Ross
>>>>
>>>>
>>>>
>>>> On 6/20/14, 3:26 AM, "Pierpaolo Cacciotto"
>>>> <pierpaolo.cacciotto.dsf.unica.it> wrote:
>>>>
>>>> >Dear Ben,
>>>> >
>>>> >yes that's the only change I made in the code and I can say that, at
>>>> least
>>>> >for my system, it is working.
>>>> >
>>>> >About my system, I should say that is an in-vacuo equilibration of a
>>>> >protein inserted in a pre-equilibrated POPE membrane.
>>>> >After the insertion there were gaps between protein and membrane so
>>>> that
>>>> I
>>>> >could not add water molecules "safely".
>>>> >Therefore I needed to equilibrate the system in-vacuo before adding
>>>> water
>>>> >but without this change in the code the system didn't equilibrate
>>>>well
>>>> (in
>>>> >particular, the membrane thickness changed). Now, with this change it
>>>> >seems fine to me.
>>>> >
>>>> >
>>>> >Regards,
>>>> >Pierpaolo
>>>> >
>>>> >
>>>> >
>>>> >
>>>> >> .Ylenia: The only option that is currently supported in Amber is to
>>>> use
>>>> >> distance restraints like in umbrella sampling. Note that group
>>>> >>restraints
>>>> >> are currently not available in the GPU-accelerated pmemd.cuda code,
>>>> so
>>>> >>you
>>>> >> are limited to pmemd and pmemd.MPI.
>>>> >>
>>>> >> .Brian: It is correct that SHAKE is only available for hydrogen
>>>> bonds
>>>> >>or a
>>>> >> set of bonds between atoms. It is currently not possible to
>>>> constrain
>>>> >> molecules or groups of molecules with SHAKE.
>>>> >>
>>>> >> .Pierpaolo: Is that all that you changed in the source code? I
>>>>think
>>>> >>this
>>>> >> should be added as an official option to Amber so that any
>>>> combination
>>>> >>of
>>>> >> X Y Z restraints can be selected.
>>>> >>
>>>> >> All the best,
>>>> >> Ben Madej
>>>> >> UCSD Chemistry and Biochemistry
>>>> >> SDSC
>>>> >> ________________________________________
>>>> >> From: Pierpaolo Cacciotto [pierpaolo.cacciotto.dsf.unica.it]
>>>> >> Sent: Wednesday, June 18, 2014 10:03 AM
>>>> >> To: AMBER Mailing List
>>>> >> Subject: Re: [AMBER] Setting up z-constraint
>>>> >>
>>>> >> Dear Ylenia,
>>>> >>
>>>> >> I had almost the same problem and I partially solved it for the
>>>> pmemd
>>>> >>cpu
>>>> >> version in the following way:
>>>> >>http://archive.ambermd.org/201405/0596.html
>>>> >> (The changes in the link are now working...I just made some
>>>>mistakes
>>>> >> recompiling the program).
>>>> >>
>>>> >> Regards,
>>>> >> Pierpaolo
>>>> >>
>>>> >>
>>>> >>
>>>> >>> Dear amber users,
>>>> >>>
>>>> >>> I need help setting up my MD simulation.
>>>> >>> Basically I want to constraint my molecule on the z axis (specific
>>>> >>>depth
>>>> >>> in
>>>> >>> a phospholipid membrane) and leave it free to move in the x and y
>>>> >>> planes.
>>>> >>>
>>>> >>> Any suggestion will be greatly appreciated!!
>>>> >>>
>>>> >>> Ylenia
>>>> >>>
>>>> >>> --
>>>> >>> *Ylenia Cau, PhD Student*
>>>> >>> *Department of Biotechnology, chemistry and pharmaceutical
>>>> sciences, *
>>>> >>> *University of Siena, Via a Moro 2, 53100 Siena, Italy*
>>>> >>> _______________________________________________
>>>> >>> AMBER mailing list
>>>> >>> AMBER.ambermd.org
>>>> >>> http://lists.ambermd.org/mailman/listinfo/amber
>>>> >>>
>>>> >>>
>>>> >>
>>>> >>
>>>> >> --
>>>> >> Pierpaolo Cacciotto - PhD Student
>>>> >> Department of Physics
>>>> >> University of Cagliari
>>>> >>
>>>> >> Cittadella Universitaria
>>>> >> 09042 Monserrato - Italy
>>>> >>
>>>> >> Phone: +39 070 675 4839
>>>> >> E-mail: pierpaolo.cacciotto.dsf.unica.it
>>>> >>
>>>> >>
>>>> >> _______________________________________________
>>>> >> 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
>>>> >>
>>>> >>
>>>> >
>>>> >
>>>> >--
>>>> >Pierpaolo Cacciotto - PhD Student
>>>> >Department of Physics
>>>> >University of Cagliari
>>>> >
>>>> >Cittadella Universitaria
>>>> >09042 Monserrato - Italy
>>>> >
>>>> >Phone: +39 070 675 4839
>>>> >E-mail: pierpaolo.cacciotto.dsf.unica.it
>>>> >
>>>> >
>>>> >_______________________________________________
>>>> >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
>>
>>
>
>
>--
>Pierpaolo Cacciotto - PhD Student
>Department of Physics
>University of Cagliari
>
>Cittadella Universitaria
>09042 Monserrato - Italy
>
>Phone: +39 070 675 4839
>E-mail: pierpaolo.cacciotto.dsf.unica.it
>
>
>_______________________________________________
>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 Mon Jun 23 2014 - 08:30:03 PDT
