Re: CCL:explicit connectivity for PDB structure in Gaussian
- From: Eric Bennett <ericb[at]pobox.com>
- Subject: Re: CCL:explicit connectivity for PDB structure in
Gaussian
- Date: Tue, 14 Dec 2004 11:06:35 -0600
Lily wrote:
I read a PDB structure into Gaussian and found that the bond order for
binding ligand is not right. I guess it is because the initial structure
is not good enough for proximity criteria in Gaussian. Then I wonder how
to make sure the bond order for both enzyme and substrate is right and
since it is a big system, it is hard to check manually.
I really appreciate if you could let me how to get the right
connectivity either changing parameter set up or using other package.
(Prefer free available)
Thanks a lot,
Lily
It appears that OpenBabel (free) will correctly assign
bond orders to
protein residues if you use it to convert your PDB file to mol2 format
(which stores bond order information in the .mol2 file):
babel -ipdb file.pdb -omol2 file.mol2
http://openbabel.sourceforge.net/
I don't use Gaussian but I presume it's smart enough to use the bond
order information stored in a mol2 file.
Commercial packages like Insight and Maestro can also assign protein
bond orders, if you have access to either of those and don't feel like
compiling OpenBabel... Maestro is nice because if it doesn't recognize
a
particular molecular fragment, it colors it differently from the rest
of
the molecule so you know to check the bond orders.
But unless you are screening a large library of structures, it's best
to
always check your ligands manually, no matter how large they are and
even if you are using a program which usually gets the bond orders
right. You might be surprised at the number of ligands in the PDB
that,
upon close inspection (sometimes it doesn't even require close
inspection!), appear to have been refined incorrectly because (a) they
have obvious distortions from the expected geometry, (b) the x-ray
resolution is low enough that such deviations should be assumed to be
errors unless otherwise explained and (c) the authors don't bother to
comment on the origin of the deviations in their paper.
You'll probably want to manually inspect certain types of residues
anyway, to make sure the protonation states make sense given available
hydrogen bonding partners, and to correct for slip-ups like glutamines
and asparagines that need to have their side chain ends flipped to make
sensible h-bonds. Some commercial packages automate this task, but I
would still check the active site area to make sure the results are
sensible.
--
Eric Bennett, Center for Drug Design, U of Minnesota