Thanks for the help, I'm actually not dealing with a transformation that changes the net charge on the system - which does open up these issues you describe.
My protein keeps the same charge in both end lambda states but I was thrown off by the setup of the TI calcs in AMBER14 with the combined PDB, where you duplicate the system (so 2 proteins, one with the mutated residue). In leap it sees both, so complains of a total charge that is double what a single protein carries and would need double the ions. I thought that the alchemical implementation would be only seeing one full system during simulation, but was not sure there was no other way it could be setting up the end states from the combined pdb file. Making the merged topology files with the parmed.py utility makes that more clear.
-----Original Message-----
From: Jason Swails [mailto:jason.swails.gmail.com]
Sent: Thursday, September 25, 2014 7:21 AM
To: AMBER Mailing List
Subject: Re: [AMBER] neutralizing systems for AMBER TI
On Sep 24, 2014, at 7:02 PM, Lawrenz, Morgan <mlawrenz.amgen.com> wrote:
> Hi all,
> For setting up a residue mutation relative free energy calc in AMBER14 with TI/FEP, I need to neutralize my system.
> The setup in leap requires a combined file with the 2 full proteins, and I wasn't sure whether I use the real net charge on a single protein for adding counter ions or to use the combined charge? I'd think just ions for a single protein, despite leap complaining that the combo structure and topology file isn't neutralized, but wanted to confirm.
Assuming of course you are running an explicit solvent calculation with plans to use PME, I would say pick one state to "neutralize" and do that (I typically add enough ions to neutralize and then some more to model a solution with some salt).
Having one state carry a net charge of +/-1 is not really a problem for PME simulations due to the fact that most (all?) implementations use what amounts to a uniform, net neutralizing plasma smeared across the entire unit cell. Since most thermodynamic cycles require you to do some sort of reference calculation, as long as you do the same charge transformation in both legs of the cycle, you should be fine.
There has been some discussion regarding net charge artifacts on this list recently (along with some citations), so if you're concerned you can try to find that discussion. The summary as I recall is that the artifacts are most pronounced for small unit cells.
HTH,
Jason
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
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 Thu Sep 25 2014 - 09:30:02 PDT
