Thanks a lot for your comments, David, they are very usefull. I will build next with ChemDraw12 and convert 2D to 3D, see if that is giving better geometries. What do you use for building unusual pieces?
Krisztina
--------------------------------------------
On Sat, 6/22/13, David A Case <case.biomaps.rutgers.edu> wrote:
Subject: Re: [AMBER] GAFF atom type for CO
To: "AMBER Mailing List" <amber.ambermd.org>
Date: Saturday, June 22, 2013, 6:59 PM
On Sat, Jun 22, 2013, Krisztina Feher
wrote:
> Sorry, that was my first failed attempt. I attached a
correct pdb
> file. The CO carbon is assiged as "c2" by divcon and as
"c1" by sqm in
> the output .mol2 file.
This is still a very odd structure. The NZ nitrogen in
one residue
(DAP) is charged (NH3), but the equivalent position on the
neighboring
residue (IGC) is not (NH2). The "alanine" is
terminated as an aldehyde
residue. It would be better to use ACE and NME capping
groups, as
Francois suggested.
The IGC residue is badly formed: there are two atoms named
HT1, and then
the "HT2" proton listed in the pdb file as part of the
Alanine residue is
actually part of IGC. If I remove the duplicate atom
names and put the
dangling HT2 proton in the proper residue, antechamber
assigns "correct"
atom types, with all the carbonyl carbons as "c:.
Notes:
1. As noted before, divcon vs sqm is irrelevant. You
can save a lot of time
by leaving off the "-c bcc" flag, while experimenting to get
your molecule
put together correctly. Getting "bad" atom types
sometimes reflects a bug
in antechamber, but more often (as here) reflects errors in
the input
structure.
2. Accerlrys Discovery Studio really seems to be putting out
bad structures.
[I can't tell if this is the fault of the program, or of the
way you used the
program....] You'll have to spend time examining things very
carefully by
hand. In particular, your structure for IGC is
unlikely to be the one you
really want. For charge and parameter development, I'd
suggest tackling IGC
and DAP in separate steps, and not try to combine two
non-standard residues
into a single structure.
3. It is also true that antechamber doesn't give
especially good informative
messages when it gets into trouble. (Requiring "good"
input is a common
feature of much academic software). In Junmei's
defense, it would take a lot
of chemical intuition (encoded as artificial intelligence, I
guess) to figure
out what was wrong with the structure you provided.
Just continue to be
careful, and look closely at all your outputs.
...dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Jun 22 2013 - 11:30:02 PDT
