Re: [AMBER] Free energy calculation of ligand binding energy: Thermodynamic integration - pmemd - single topology - vdw

From: <vladimir.palivec.marge.uochb.cas.cz>
Date: Sat, 25 Apr 2015 13:58:53 +0200

Hi,

 thanks, now the simulations converge much faster! I also stumbled on
another question: Is it possible to perform MBAR analysis of simulations
using pyMBAR? I tried the logdvdl=1 option, however, I seems to me it is
not possible to use pyMBAR - as it only supports TI and TI-CUBIC. I also
tried the mdout2pymbar.pl script, however, it look like it is witten for
some old version of pyMBAR. As it seems to me now, it is only possible to
analyze gromacs data format(dhdl.xvg).

Thanks for any suggestions,
Vlada

> Hi,
>
> What you are trying do is a relative free energy simulation. Your
> softcore masks are identical with the TI region which means that you are
> running purely dual-topology, i.e. you have two independent copies. No
> wonder that they drift apart.
>
> What you probably should do is to find a suitable mapping between two
> ligands. Programs like FEW (part of the AmberTools) or FESetup can help
> you doing this. The mapped region would than be treated via a single
> topology which implies also a single coordinate set for those atoms. This
> provides you with a suitable "anchor" which a) ties your ligands together
> and b) does not have completely vanishing end states similar as you would
> have in an absolute free energy simulation (you could consider a
> simulation as you have attempted to be two absolute calculation
> simultaneously but only getting half of the results).
>
> Cheers,
> Hannes.
>
> ________________________________________
> From: vladimir.palivec.marge.uochb.cas.cz
> [vladimir.palivec.marge.uochb.cas.cz]
> Sent: 23 April 2015 16:26
> To: amber.ambermd.org
> Subject: [AMBER] Free energy calculation of ligand binding energy:
> Thermodynamic integration - pmemd - single topology - vdw
>
> Hello again,
>
> I am trying to learn free energy calculations, however, it is much harder
> than performing simple MD simulation. I would like to ask about
> alchemistical calculation of mutating one ligand to another, especially
> about vdw change step. When I mutated ligand1->ligand2 at lambda=0.25 a
> got this result
>
> http://postimg.org/image/adoq76m1r/
>
> which shows some strange thing happening at a step of about 2000. So I
> checked trajectory and this is what I see :
>
> http://postimg.org/image/6ejpdhvub/
>
> this image shows distance between ligand1 and ligand2. I am confused how
> the DV/DL is beeing calculated and whether the calculation setup is right
> - as clearly both ligands are interacting between each other.
>
> I have used following setup:
>
> icfe=1, clambda = 0.25,
> ifsc=1,
> timask1=':2', scmask1=':2',
> timask2=':1', scmask2=':1',
> crgmask=':1,2',
>
> So my questions are following: What is the right way to calculate the vdw
> change step? Is it ok to change whole ligands ( I am changing phenol to
> serotonin for example)? Isn't it a problem - the result I got ?
>
> Thank you very much for any advice!
> Vlada
>
>
>
> _______________________________________________
> 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
Received on Sat Apr 25 2015 - 05:00:03 PDT
Custom Search