Hi James,
It sounds like your membrane is not initially equilibrated. The box size likely shrinks a lot when you switch to NPT - this is normal for proteins embedded in membranes due to the packing not being ideal when building the membrane.
Firstly I would switch to NPT as soon as possible - if you stick with NVT too long you may start to make vacuum bubbles which can make the density equilibration much more difficult.
Next, yes I would set skinnb higher during the initial NPT run, when you see the skinnb crash (or use the CPU code for this section of the run). Once your membrane has equilibrated you can set skinnb back to the default (just remove it from the ewald namelist). Setting it large does not affect results but will slow the simulation down quite a bit so it's best to only have it set high when you need it.
Note, you should also monitor your area per lipid and bilayer thickness to ensure that things have equilibrated. Take a look at the following tutorial for an example heating and equilibration protocol.
http://ambermd.org/tutorials/advanced/tutorial16/
All the best
Ross
> On Feb 10, 2015, at 7:44 AM, James Starlight <jmsstarlight.gmail.com> wrote:
>
> Dear Amber Users!
>
>
> This time I'm in charge with running of several identical systems consisted
> of membrane receptor within the membrane (the difference actually only in
> the ligands within the cavity of the receptor). For those simulations I'm
> using my equilibration protocol consisted of the NVT 2-step heating with
> position restrains applied on the receptor's atoms, and further 2 steps of
> the NPT equilibration w/o of any restrains. After this I run production run
> which I use for the MMGBSA analysis. From this (md run) typically first
> 15-20ns is considered as the equilibration and the last 50-70 is used for
> the mmgbsa. Approximately in 2 of the 10 simulations of the identical
> systems I got crash during those first 15-20ns period with the skinnb error
> which associated with the rapid alterations in the box dimensions during
> the lipids equilibration around the receptor. Typically I modify this term
> only during last stage in NPT equilibration phase but not in the beginning
> of the stage where actually I got the crash
> #npt
>
> &ewald
>
> skinnb=5, ! Increase skinnb to avoid skinnb errors
>
>
> After this if I run new simulation from the last checkpoint where I got the
> crash- all go fine.
>
> So the question whether I should to modify this term also during beginning
> of the production run (e.g use skinnb=3) to avoid the crash or there are
> any other solutions here?
>
>
> Thanks for help,
>
>
> James
> _______________________________________________
> 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 Tue Feb 10 2015 - 09:30:02 PST
