AMBER Archive (2007)

Subject: Re: AMBER: coordinate changes using ptraj

From: David A. Case (case_at_scripps.edu)
Date: Mon Mar 05 2007 - 10:41:42 CST

On Sat, Mar 03, 2007, Mike Summers wrote:

> I have a general procedural question. I have used amber
> to generate RNA NMR structures using Na+ and explicit water.
> As I understand it, final minimization should probably be
> done without the water. I therefore would like to
> use ptraj to strip out the water after doing the MD
> calculations and save the restart file that does not
> have water. I would then use the water-less restart file
> for final minimizations in vacuo.

I think most people doing explicit solvent refinements never go the stage of
a "final refinement", but rather cool the system down to room temperature, and
save samples from that distribution. Because you have thermal energy,
deviations from the NMR constraints will generally be somewhat larger than
with fully minimized structures, but that is generally considered more of an
apparent than a real problem.

As Carlos pointed out, more conventional looking refinements that use the
generalized Born solvent model are also possible, and there you could do a
minimization.

>
> The potential problem is that, if the coordinates are
> modified by the strip command, I think this will lead to major
> RDC violations. Is it possible to strip the water and Na+ atoms
> without changing the coordinates?

I don't think the coordinates should change when you strip waters....

>
> I'm having what I think is a related problem in a calculation
> I'm doing now, in which I use a reference structure to hold
> atoms rigid during an MD (in water, Na+) simulation. After
> doing the MD, if I use ptraj to strip the water and save a
> water-less restart file, the restart file LOOKS good when I
> view it in pymol, but when I use it to start a restrained
> minimization (again, using a reference structure to fix
> some of the atoms), the amber energies are off the charts...

Is it the restraint energy that is "off the charts"? There is an intrinsic
problem with using restraints and constant pressure simulations together.
The volume of the system is changed by scaling the coordinates, but the
restraint coordinates are not scaled in the same way. To do a restart, you
may need to restrain to the starting coordinates of that run, not to the
orignal coordinates. It may also be that "wrapping" the coordinates could be
causing this problem. But mostly, I'm guessing here, since I don't have
enough information to really help. You might want to post some files (or send
them to me privately) so that we can narrow down the problem. I don't have
much personal experience with explicit solvent refinement, but there are a
number of pitfalls (as you have found!) that we should be able to work
through.

...dave

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu