AMBER Archive (2008)

Subject: Re: AMBER: pmemd iwrap trouble

From: Robert Duke (rduke_at_email.unc.edu)
Date: Tue Aug 26 2008 - 11:59:34 CDT


Hi Tom,
Thanks, that was my understanding that the imaging in sander and pmemd was
the same, and "equivalent" to a truncated octahedron, but that the image
would not look correct in something like vmd. I was hoping to get Tom D. to
comment on this since he did the code. I don't use truncated octahedrons
that much myself, but have had 0 problems with other imaging, aside from
problems with not being quite sure what to do in ptraj sometimes, and also
ptraj understandably can't keep up with box size changes associated with a
constant pressure simulation (so wrapping a whole trajectory rather than a
frame really blows up; one might expect that...). Anyway, thanks for the
input. I am going to confirm that my code is identical to sander in amber
10 with regard to wrapping, but that is what I expect. There IS a known
problem with the amber 9 release of pmemd, for which a patch was released,
so some folks with an unpatched version of 9 may see a few molecules not
wrap correctly (so do the patch...).
Regards - Bob

----- Original Message -----
From: "Thomas Cheatham III" <tec3_at_utah.edu>
To: <amber_at_scripps.edu>
Sent: Tuesday, August 26, 2008 12:48 PM
Subject: Re: AMBER: pmemd iwrap trouble

>
>> I get exactly the same results from pmemd9, pmemd10, and sander10 in
>> regard the wrapping (iwrap=1); 2 of my subunits are displaced
>> (regardless of ioutfm=1 or 0). However, by using the following ptraj
>> script (instead of iwrap) it results in a truncated octahedral fitted
>> nicely around the protein complex (as expected):
>>
>> center :1-5332
>> image center familiar
>> trajout reimaged.rst restart
>>
>> However, I see no trouble when I use solvatebox instead of solvateoct
>> (for both sander and pmemd). Unfortunately I cant go back using
>> solvatebox at this time, after 2 months of simulations..
>
> I've been following this thread and think that it is just a
> misunderstanding and that nothing is wrong with the imaging, it just
> doesn't look like you would expect. sander/pmemd image in triclinic space
> so for a protein this looks like a slanted box with the protein
> potentially sticking out...
>
> __PPP__
> / PPP /
> /___PPP/
>
> ...rather than the more "familiar" truncated octahedron shape. To verify
> this, look at the ptraj imaged trajectory without the familiar keyword;
> this will look like what you are seeing from sander/pmemd and they are
> equivalent, albeit different ways of looking at things.
>
> There is also a way in pmemd to generate images in alternative unit cells,
>
> image xoffset 1.0 yoffset 0.0 zoffset 0.0
>
> which you can do for each direction and then "see" the whole packing in
> the unit cell.
>
> If you are still concerned, you could also try a single point energy
> minimization on both the "familiar" and alternative restrt file to verify
> the energies are similar.
>
> imaging in ptraj is not broken (now) to the best of my knowledge; there
> was a brief problem with AmberTools 1.0 that lead to errors in the box
> size, but this was working in earlier versions and is working in current
> versions.
>
> --tom
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
> to majordomo_at_scripps.edu
>

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
      to majordomo_at_scripps.edu