On Tue, Feb 3, 2015 at 1:30 PM, Kenneth Huang <kennethneltharion.gmail.com>
wrote:
> Jason,
>
> It turns out it was very probably the numbering issue from applying the
> strip mask- when I change the mask to -s ':WAT,:Cl-,:1-1044' -n ':1-522'
> to account for the now absent residues, it worked without any problems. I
> hadn't realized that parmed.py would do the sequential renumbering of the
> residues, and instead just pull them straight out from the prmtop.
>
To shift a little of the blame (credit?) away from ParmEd here -- parmed
never *renumbers* anything. The sequential numbering of atoms and residues
inside the Amber topology is implicit by virtue of an explicit numbering
system being absent. So the question becomes one of order-of-operations.
Does ParmEd strip the strip_mask first, and then generate the ligand and
receptor topologies from the complex topology? Or does it generate all 3
from the solvated topology?
If it does the former, you get the "renumbering" effect. If it does the
latter, then you don't. If the strip_mask is always at the end, then this
distinction doesn't matter (since the atoms that remain have the same
implicit number in both the solvated and dry complex prmtops). Since
strip_mask is not at the end in your case, this distinction *does* matter.
It turns out this is loosely documented in the --help output:
-m RECEPTOR_MASK, --receptor-mask=RECEPTOR_MASK
Amber mask of atoms needed to be stripped from
COMPLEX
to create RECEPTOR. Cannot specify with -n/--ligand-
mask
-n LIGAND_MASK, --ligand-mask=LIGAND_MASK
Amber mask of atoms needed to be stripped from
COMPLEX
to create LIGAND. Cannot specify with -m/--receptor-
mask
So while it's nominally "documented", it's certainly far from obvious. In
retrospect I think the other choice (to have all masks apply to the same
topology file) may have been a better one, or at least more intuitive to
more people. But it's not enough of an improvement IMO to change its
behavior.
That said, ParmEd should never choke on an IndexError -- that is clearly a
bug that needs to be addressed. Chances are it's fixed in the current
developer's version (in which case I will probably punt on the fix until
AmberTools 15 comes out :).
In any case, thanks for the report and the good question. I will try to
clarify the docs a bit.
All the best,
Jason
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Feb 03 2015 - 12:30:03 PST
