I rest my case: BOGUS!!!
That paper gets the award for stating the bleeding obvious. How people were
able to publish such things in 2004 is beyond me. I thought this issue had
been more or less put to bed when PME came out in the 90s.
Their conclusion is pretty simple. If you use stupid spherical truncation
methods you ("quite obviously!") see issues due to that truncation and
variation in behavior as a function of cutoff. Now why on earth ANYBODY
would still be doing such things 20 years after the solution (PME) was
published I have absolutely no idea. As they show in that paper, PME fixes
it (and is a damn site faster to boot) so why do they bother with the rest
of the tests? Oh to get a publication, of course. :-(
There is a reason that AMBER's PMEMD does NOT support simple spherical
cutoffs. Use PME with a 8 to 10 angstrom cutoff and you should be fine. We
have checked and the cutoff value once above 8 has little to no difference
on the applied surface tension. You might see very minor differences due to
truncation of VDW terms at the cutoff but these are very minor.
Short answer: The person at the charmm workshop was most likely misinformed
or had not kept up with the current recommended approaches for MD
simulations. Papers like this one don't help unfortunately.
Now that said, it may be necessary to adjust the constant surface tension
term as a function of the number of lipids in your bi-layer although I would
expect that to converge as the number of lipids increases.
All the best
Ross
> -----Original Message-----
> From: Albert [mailto:mailmd2011.gmail.com]
> Sent: Friday, June 15, 2012 3:36 AM
> To: AMBER Mailing List
> Subject: Re: [AMBER] R: strange POPC under lipids 11 FF
>
> I didn't remember the details now and I have to check my notes.
> However,
> I found a similar paper may concerned on this issue:
>
> http://pubs.acs.org/doi/full/10.1021/jp048969n
>
>
>
> On 06/15/2012 12:38 PM, Ross Walker wrote:
> > Did they explain why? Seems bogus to me.
> >
> >
> >
> > On Jun 15, 2012, at 11:28, Albert<mailmd2011.gmail.com> wrote:
> >
> >> On 06/15/2012 12:02 PM, Ross Walker wrote:
> >>>> Moreover, the surface tension values probably should also have
> >>>>> something
> >>>>> to do with the cutoff we use..... I don't know whether Ross have
> any
> >>>>> comments on this aspects.
> >>> No it should NOT depend on the cutoff. Why would you think it
> should depend
> >>> on the cutoff?
> >>>
> >>> If it does then your cutoff is TOO SMALL, period!
> >>>
> >>> I suggest you check things carefully. Ultimately if you want to
> publish the
> >>> work you will need to defend the surface tension you used so you
> might as
> >>> well run a few hundred nanoseconds for the system you want to run
> with a
> >>> range of surface tension values and then you'll be able to provide
> a
> >>> justification of the value you used if a reviewer asks.
> >> that's strange, because in recent CAHRMM workshop, the speaker
> emphasize
> >> this issue and they state that the non-bonded cut off should
> influence
> >> the surface tension.....
> >>
> >> _______________________________________________
> >> 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
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Jun 15 2012 - 16:00:05 PDT
