AMBER Archive (2008)

Subject: Re: AMBER: pmemd iwrap trouble

From: Robert Duke (rduke_at_email.unc.edu)
Date: Thu Aug 28 2008 - 17:05:05 CDT

Hi Lars and Tom,
I also have been looking at the coordinates and running a few simple tests
(and I concur it is an "ambitious" effort). It also looks to me like this
is the typical "subunit separation" problem with big solute complexes that
are not completely covalently bonded. Confirming that everything is fine is
just a bit more complex due to all the machinations involved in getting
things from the triclinic to the truncated octahedral unit cell. I did do
one simple test, which was to run a 0.00001 fsec single step with and
without wrapping, and then run another 1fsec step without wrapping, and look
at all the energies. They match to 8 digits for everything except vdw,
which barely misses 8 digits. If the wrapping was hosed, it would be rather
unlikely for the delta's in energies to be this small, I would think. I
will probably also do some of this in sander just for grins; I expect that
things will be pretty much the same.
Best Regards - Bob

----- Original Message -----
From: "Thomas Cheatham III" <tec3_at_utah.edu>
To: <amber_at_scripps.edu>
Sent: Thursday, August 28, 2008 5:47 PM
Subject: Re: AMBER: pmemd iwrap trouble

> Good energies or not, two of my subunits do still not fit into the
> rest of the complex properly. Total rmsd of 70Å when I do not reimage
> the wrapped restart file, and 1.6Å when I do reimage the wrapped
> restart file (with ptraj).
>
> You mentioned that my subunit complex could be spanning the primary
> unit cell, and that two subunits are therefore individually
> wrapped(?). This makes sense, but should it not be possible to reimage
> this with ptraj and not get an rmsd of 1,6Å? I can clearly see that
> two subunits are shifted.

Lars sent me his coordinates and I can verify that the coordinates before
and after imaging are essentially equivalent with the exception that one
of the subunits is in an adjacent periodic image cell after PMEMD imaging
and not ptraj imaging. The difference likely results due to the use of
the "center" command in ptraj to center the protein prior to imaging. I
can also verify that this is an incredibly large system!

Lars, if you do not center before imaging in ptraj likely you will get the
same thing as you saw with PMEMD. What is happening is that the whole
protein complex is moving in the box (as expected). One of the subunits
has its center of mass in an adjacent unit cell so when imaged moves to
the other side of the unit cell. There is no way to easily prevent this
except by (a) restraining the whole complex to stay at the center of the
box or by (b) altering the prmtop to treat all the protein as a single
molecule. You can however, most likely, through clever imaging in ptraj
"fix" this after the fact likely using the center/image commands.

So, I think you can continue with your PMEMD wrapping (i.e. restart from
the b4 restrt and continue running turning on IWRAP=1) and then later use
ptraj to put the complex back together again. [e-mail me offline if you
want to have the prmtop modified so that the protein will never
"dissociate" to periodic images...]

--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