RE: Re: alchemical free energy simulations with parmtop

From: hannes.loeffler_at_stfc.ac.uk
Date: Thu Dec 04 2014 - 11:52:49 CST

Hi Jim,

many thanks. One other point to consider may be that the code checks if alchLambda2 is set. I think for TI calculations this is not needed.

I have made some tests in the meantime computing relative hydration free energies and I believe that this _may_ work. The reason why I am not fully sure yet is how the particular alchemical implementation works in NAMD. With NAMD I can only define disappearing or reappearing atoms but not direct scaling of one atom type into another (would require scaling of all bonded terms). I can "mimic" this with a phase-in-phase-out scheme, i.e. annihilate one atom type and recreate it as that of the final state. But the practical problem is the charge as I may end up with a non-zero net charge in one of the partitions. A way around this is to go "full" dual topology i.e. annihilate the first state and exnihilate the other... I guess this all works nicely with the example in the manual and the CHARMM force field which defines IIRC charge neutral groups (or integral charged if necessary). I am dealing, however, with "arbitrarily" parameterised organic ligands.

Many thanks again,
Hannes.

________________________________________
From: Jim Phillips [jim_at_ks.uiuc.edu]
Sent: 04 December 2014 17:09
To: namd-l_at_ks.uiuc.edu; Loeffler, Hannes (STFC,DL,SC)
Subject: RE: namd-l: Re: alchemical free energy simulations with parmtop

Hi again,

>From checking my old emails it appears I added the test to disallow
alchemical simulations with AMBER parameters because one of our free
energy gurus unfamiliar with the details of the code thought there was
likely an issue. On further inspection AMBER parameters should work fine,
so I've removed that check but added some warnings for VDW switching
options (off, force, or Martini) that will not be used for alchemical
interactions (but will apply to interactions not involving product or
reactant atoms). These changes are in the 2014-12-04 nightly build.

Thanks for pointing this out.

Jim

On Tue, 2 Dec 2014, hannes.loeffler_at_stfc.ac.uk wrote:

> Hi Jim,
>
> many thanks for your reply.
>
> I am not sure though if I fully understand what you are saying besides that I should give it a try. I see that this was changed in rev 1.1439 of src/SimParameters.C (by 'jim', is that you?) and naturally I was wondering if there is a deeper reason for that.
>
> The most obvious question to me is if I have to modify the parmtop file to get this working. I have found the code which deletes all bonded terms between the appearing and disappearing atoms. I have not yet found the bit which does this for the non-bonded parameters although I should be able to do just that by adding appropriate exclusions in the parmtop file.
>
> Can namd do alchemical simulations in vacuo?
>
> Cheers,
> Hannes.
>
> ________________________________________
> From: Jim Phillips [jim_at_ks.uiuc.edu]
> Sent: 02 December 2014 18:06
> To: namd-l_at_ks.uiuc.edu; Loeffler, Hannes (STFC,DL,SC)
> Subject: Re: namd-l: Re: alchemical free energy simulations with parmtop
>
> There is nothing obvious to me. I can understand certain switching
> options not being supported, so perhaps AMBER has a different default.
>
> You can certainly remove the check and try.
>
> Jim
>
>
> On Mon, 1 Dec 2014, Hannes Loeffler wrote:
>
>> Right. So namd explicitly disallows alchemical simulations when amber
>> is on. The error message is a bit weird though as it says that "AMBER
>> parameters" are not supported. A particular parameter set in itself can
>> hardly be the reason as it should be possible to introduce those trough
>> the PSF or Gromacs .top route (although the particular functional form
>> of the AMBER force fields may not be fully supported). I can't see
>> any safeguards against this in the code.
>>
>> So, the question: why would the AMBER _force field_ not be supported?
>> Apparently there are a few differences in the functional forms of the
>> CHARMM and AMBER force fields which lack support for instance.
>>
>> Cheers,
>> Hannes.
>>
>>
>> On Tue, 17 Jun 2014 10:35:59 +0100
>> Hannes Loeffler <Hannes.Loeffler_at_stfc.ac.uk> wrote:
>>
>>> Hi,
>>>
>>> I am just looking through a few tutorials on how to do alchemical
>>> relative free energy calculations and see that they are all based on
>>> the PSF topology file format. I wonder if this is a requirement with
>>> namd or a AMBER parmtop file could be used instead.
>>>
>>> From what I understand there could be problems with the force fields
>>> terms in the parmtop because the way it works in namd is to create the
>>> morph pair as a single residue (and describe vanishing/appearing atoms
>>> in an associated PDB file). But is it possible with a parmtop
>>> provided the parmtop is created/manipulated accordingly? I guess
>>> this all boils down to how namd generates the force field terms
>>> internally for the two mutated molecules/fragments.
>>>
>>> Many thanks,
>>> Hannes.
>>
>> --
>> Scanned by iCritical.
>>
> --
> Scanned by iCritical.
>

-- 
Scanned by iCritical.

This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:23:05 CST