Re: [AMBER] some problems of restart MD

From: Jason Swails <>
Date: Thu, 2 Dec 2010 09:57:38 -0500


2010/12/2 cwt <>

> I have read the Amber10 manual and some papers of constant pH MD method, it
> seems that there's no need to heat and equilibrate the system. And the input
> file parameters are set as the page 135 of Amber10 manual. If I

I assure you this is a good idea; all MD simulations on real systems (i.e.
proteins, etc) require careful heating and equilibration to make sure that
you're sampling from an equilibrium ensemble. Constant pH MD simulations
are no different, and is in fact more susceptible to unraveling than normal
MD due to the fact that you're changing the net charge on the system

There are certain input parameters that you definitely want to keep the
same, such as igb, saltcon, tol, cut, scee (although 1.2 is the default),
ntb, ntc, and ntf (and nrespa should at least be such that even slow-varying
terms are updated each time you take a MC move, though 1 is recommended).
These need to be kept the same because they actually potentially affect the
potential energies of the reference compounds, which has been parameterized
with these parameters beforehand. However, the only affect that the
thermostat may have is to alter the distributions of the sampled
conformations of the reference compound. However, the reference compound is
so small, and the simulations are run for so long, that this really has no
effect (I've verified it, and it's easy to verify it yourself). Therefore,
I think that the benefits conferred by the langevin thermostat are well
worth the change.

> run for 50ns without stop, it will be too long to wait for it. Because the
> software was not setuped by me, I don't make sure that whether it has fixed
> all bugs. And after bug fixed, does it need a recompilation?

Yes it does.

> Thank you for your attention!

Good luck!

> At 2010-12-02 12:43:59,"Jason Swails" <> wrote:
> >A couple comments here... First, make sure you have applied all bug fixes,
> as there were some fairly serious issues with running constant pH MD with
> multiple titratable residues. Second, you should expect larger fluctuations
> in constant pH simulations, since you're abruptly changing charges when
> protonation moves are accepted. Make sure you carefully heat and
> equilibrate your system with restraints before you begin to run production
> dynamics. One other comment: you should probably switch from the Berendsen
> thermostat (ntt=1) to the Langevin thermostat (ntt=3), and change tautp to
> gamma_ln (see the manual for more details). The thermostat has no effect on
> the state energies, so it won't affect sampled protonation states.
> >
> >Hope this helps,
> >Jason
> >
> >--
> >Jason Swails
> >Quantum Theory Project,
> >University of Florida
> >Ph.D. Graduate Student
> >352-392-4032
> >
> >On Dec 1, 2010, at 9:21 PM, cwt <> wrote:
> >
> >> Dear all,
> >> I have done the constant pH MD of Amber for 1ns. The result is good.
> Then I try to run for 50ns, which restarts from the coordinates and
> velocities of the 1ns. But I the energy changes very much, and the
> structure is unstable. Following is the .in file of mine:
> >> &cntrl
> >> imin=0, irest=1, ntx=7, ntb=0,
> >> igb=2, saltcon=0.1, ntpr=1000, ntwx=1000,
> >> ntt=1, tautp=2.0,
> >> tempi=300.0, temp0=300.0,
> >> scee=1.2, nrespa=1,
> >> ntc=2, ntf=2, tol=0.000001,
> >> nstlim=2500000, dt=0.002,
> >> cut=30.0,
> >> icnstph=1, solvph=7.0, ntcnstph=3
> >> /
> >> I don't konw what causes this, By the way, my edition is Amber10.
> >> Thank you!!
> >> _______________________________________________
> >> AMBER mailing list
> >>
> >>
> >
> >_______________________________________________
> >AMBER mailing list
> >
> >
> _______________________________________________
> AMBER mailing list

Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
AMBER mailing list
Received on Thu Dec 02 2010 - 07:00:03 PST
Custom Search