Re: [AMBER] Decoupling of the restrained ligand using Thermodynamic Integration in Amber ?

From: <steinbrt.rci.rutgers.edu>
Date: Thu, 10 May 2012 03:33:34 -0400 (EDT)

Hi Marek,

I never tried what you describe. It is in principle a good idea, but would
produce enormous sampling problems:

The non-interacting, non-restrained ligand has the entire box as accesible
phase space, so for a converged simulation it would have to repeatedly
cover every spot therein. This is almost guaranteed not to happen with
even the smallest restraints, so in terms of phase space overlap between
initial and final state, this is about the worst case possible in a free
energy simulation. So, yes you can try running that, but I doubt it will
give a meaningful result.

Thomas

On Wed, May 9, 2012 11:43 am, Marek Maly wrote:
> Thanks Thomas !
>
> I have just the last particular question.
>
> In spite the fact that the very last part of the ligand decoupling from
> the receptor (using restraints):
>
> dAr
> (R)s***(L)v----->(R)s + (L)v
>
>
> (i.e. calculation of the free energy change connected with the
> "anihilation" of the restraints
> which are applied on already non-interacting (vdw/el) ligand) is possible
> to calculate analytically,
> I would like to verify it numerically by TI with this setup:
>
> V0 = Receptor + Ligand (in water) where all the ligand atoms are DUMMY
> atoms.
> Ligand is restrained in V0 with target restraints (defined in the RST
> file)
> (ifsc=0, scmask='', crgmask='')
> dvdl_norest=0
>
> V1 = Receptor + Ligand (in water) where all the ligand atoms are DUMMY
> atoms.
> Ligand is free to move in V1 (no restraints)
> (ifsc=0, scmask='', crgmask='')
> dvdl_norest=0
>
>
> Might it be correct for the given purpose ?
>
> Thank you very much again !
>
> Best wishes,
>
> Marek
>
>
>
>
>
>
> Dne Wed, 09 May 2012 17:08:52 +0200 <steinbrt.rci.rutgers.edu> napsal/-a:
>
>> Hi,
>>
>>> Regarding the scaling constant (which "preventively" scale restraints)
>>> when SC pot. are used: 1/lambda versus 1/(1-lambda) maybe it is not
>>> the "full typo" as this is probably dependent on simulated process. For
>>> the ligand "disappearing" scaling constant
>>> is 1/(1-lambda) for the opposite process (appearing) perhaps 1/lambda
>>> is
>>> used. This would correspond to the disappearing/appearing
>>> SF-potentials where in case of disappearing multiplicative constant
>>> (1-lambda) is used and in case of appearing it is just lambda.
>>
>> True, maybe that was my idea when writing it. However, I think only the
>> setup in which whole restrained ligands disappear has ever been used.
>>
>>> #1 - appearing of the restraints while ligand is fully (vdw/el)
>>> interacting with receptor during this process
>>>
>>> Here I would use this approach:
>>>
>>> V0 = Receptor + Ligand (in water)
>>> no restraint is applied in V0 ( ifsc=0, scmask='',crgmask='' )
>>> dvdl_norest=0
>>>
>>>
>>> V1 = Receptor + Ligand (in water)
>>> Ligand is restrained in V1 with target restraints (defined in the RST
>>> file) (ifsc=0, scmask='',crgmask='')
>>> dvdl_norest=0
>>
>> yes, that is how I would do it. This linearly increases the restraint
>> strength with lambda, which is what you want.
>>
>>> #2 Decoupling (vdw/el) the fully restrained ligand.
>>>
>>> Here I would use this setup:
>>>
>>> V0 = Receptor + Ligand (in water)
>>> Ligand is restrained in V0 with target restraints (defined in the RST
>>> file) (ifsc=1, scmask=':LIGAND',crgmask='')
>>> dvdl_norest=1
>>>
>>>
>>> V1 = just Receptor (in water)
>>> ( ifsc=1, scmask='',crgmask='' )
>>> dvdl_norest=1
>>>
>>
>> that is also right, but I think dvdl_norest can be left out in V1, it
>> has
>> no real effect there. It is necessary in V0. In this setup, the ligand
>> will decouple, but will still be restrained with unchanged strength.
>>
>>> Anyway the second possibility for V1 might be here:
>>>
>>> V1 = Receptor + Ligand (in water) where all the ligand atoms are DUMMY
>>> atoms.
>>> Ligand is restrained in V1 with target restraints (defined in the RST
>>> file) (ifsc=1, scmask=':LIGAND' or scmask='' ???, crgmask='')
>>> dvdl_norest=1
>>>
>>
>> No, that is not useful. Dummy atoms and softcore are mutually exclusive
>> ways to do the same thing and they cannot be mixed like this. Stick to
>> the
>> first way you suggested.
>>
>> Kind Regards,
>>
>> Thomas
>>
>> Dr. Thomas Steinbrecher
>> formerly at the
>> BioMaps Institute
>> Rutgers University
>> 610 Taylor Rd.
>> Piscataway, NJ 08854
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
>> __________ Informace od ESET NOD32 Antivirus, verze databaze 7123
>> (20120509) __________
>>
>> Tuto zpravu proveril ESET NOD32 Antivirus.
>>
>> http://www.eset.cz
>>
>>
>>
>
>
> --
> Tato zpráva byla vytvořena převratným poštovním klientem Opery:
> http://www.opera.com/mail/
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>


Dr. Thomas Steinbrecher
formerly at the
BioMaps Institute
Rutgers University
610 Taylor Rd.
Piscataway, NJ 08854

_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu May 10 2012 - 01:00:03 PDT
Custom Search