AMBER Archive (2005)

Subject: Re: AMBER: Ptraj and order of operations

From: Furse, Kristina Elisabet (kristina.e.furse_at_vanderbilt.edu)
Date: Tue Jul 12 2005 - 20:07:04 CDT

> Simple answer is yes. With the exception of trajin and trajout each
> action specificied is applied sequentially to the coordinates in the
> order listed by the user in the input file (courtesy of the AMBER
> manual).

The example input script in the ptraj section of the manual suggests to rms fit,
then center, then image, so that's what I've typically done. Just thinking
theoretically it seems safer to image first, then rms fit. That way all of the
little stray molecules go back in the main box, and then the entire box is
shifted according to the rms fit. If you do it the other way, rms fit, then
image, I guess some of the solvent will stay in the same box image, while some
may move to a neighboring one. But if everything moves the same, they should
image back the same, right?

I should say I ran into this issue because I strug together 200 restart files
spanning a 10 ns trajectory so I could get a feel for the whole thing in one
mddisplay session. I didn't use iwrap, so I have solvent that moves many virtual
boxes away from the main one by the end of the trajectory. I found that if I
used ptraj to rms fit the protein backbone first to get rid of rotational
motion, then imaged (manual-approved protocol and my usual plan), I got a
trajectory that started out "normal", but as it went along I had solvent
molecules overwhelming my active site, passing directly through protein atoms. I
know that didn't really happen in the simulation, so it must be an imaging
issue, one that is possibly exacerbated as solvent moves farther away. I noticed
that if instead, I imaged first then rms fit, I got a "normal" looking
trajectory, with well-behaved solvent. Couldn't find a simple correlation
between solvent residue number or even x,y,z position and "bad" imaging. Like I
said before, some of the solvent is imaged properly using the standard order of
operations, some is not.

Anyway, this might be a situation not many people will run into, and the easy
workaround seems to be just to image first, then rms fit. But it is weird and
stuck in my head, and I've already wasted a day of Chris Moth's time looking at
this when I though it was an mddisplay issue... Any ideas appreciated!

Kristina

>
> On 7/11/05, Furse, Kristina Elisabet <kristina.e.furse_at_vanderbilt.edu>
> wrote:
> > Hi-
> >
> > Should the order of operations matter for the rms and image commands in
> ptraj?
> > For instance, in theory should the following two scripts output the same
> > coordinates?
> >
> > trajin test.crd 2 2
> > reference test.pdb.1
> > trajout pass.pdb pdb
> > image center familiar
> > rms first out rmsCheckPass :554
> > go
> >
> > trajin test.crd 2 2
> > reference test.pdb.1
> > trajout fail.pdb pdb
> > rms first out rmsCheckFail :554
> > image center familiar
> > go
> >
> >
> > I ask b/c I have run into a situation where the above scripts give pdb's
> which
> > have coordinates for solvent and counterions that are partly identical and
> > partly different--for instance, 8 out of 12 Na+'s have identical
> coordinates in
> > the two pdb's, while 4 have different coordinates. The system is a protein
> in a
> > periodic box. I can give more detail on the problem and system, and my
> labmate
> > Chris can talk code, but I thought I'd start with the simple question in
> case
> > there was a simple answer...
> >
> > Thanks,
> > Kristina
> > -----------------------------------------------------------------------
> > The AMBER Mail Reflector
> > To post, send mail to amber_at_scripps.edu
> > To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> >
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
>

-----------------------------------------------------------------
Kristina E. Furse
Department of Chemistry
Center for Structural Biology
Vanderbilt University
Email: kristina.e.furse_at_Vanderbilt.Edu
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu