Re: AMBER: Has anyone seen a protein fall apart due to PBCs?

From: Thomas Cheatham III <>
Date: Tue, 22 Jan 2008 09:37:03 -0700 (Mountain Standard Time)

> I'm simulating with the AMBER FF99SB forcefield, and the tetrameric protein
> I've got seems to like to break into along the weaker of the two tetramer
> interfaces. To be sure, there are single point mutants of this protien known

...what pops to my mind is possible issues with equilibration at constant
pressure with positional restraints as this scales the positions of the
monomers (when NPSCAL=1) of both the real and reference coordinates...
If you initially solvated with LEaP, this tends to underestimate the
density so the system will compress slightly bringing the monomers closer
together if you run constant P with NTR=1 (restraints). When the
restraints are turned off, the monomers will slightly repulse.

Of course, this assumes that you did constant P with restraints (which I
cannot tell from the post but is consistent with standard equilibration
protocols). If you search the AMBER reflector archives for belly or
restraints and my name, there are a number of posts that show how to hand
modify the prmtop to put all the protein monomers into a single molecule
(for pressure scaling) and it would be worth checking if this is the

[If it is not due to constant P w/ restraints, my next guess would be salt
and/or hydration at the interface and suggests prudence in carefully
equilibrating w/ restraints for a long time, ~1-5 ns]


p.s. note that I do not think this would be due to periodicity artifacts;
despite the arguments of Hunenberger and Weber from your group, there is a
nice paper in JBSD by Villarreal and coworkers (JBSD23, 135 (2005)) that
suggests the artifacts in reality are minimal (< kT), as does nice work by
Bader & Chandler, Smith & Pettitt, ...
The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Wed Jan 23 2008 - 06:07:28 PST
Custom Search