From: Francesco Pietra (chiendarret_at_gmail.com)
Date: Tue Mar 22 2011 - 11:03:05 CDT

forgot the list, sorry

---------- Forwarded message ----------
From: Francesco Pietra <chiendarret_at_gmail.com>
Date: Tue, Mar 22, 2011 at 5:02 PM
Subject: Re: vmd-l: charmmgen to autopsf
To: Peter Freddolino <petefred_at_ks.uiuc.edu>

On Tue, Mar 22, 2011 at 2:37 PM, Peter Freddolino <petefred_at_ks.uiuc.edu> wrote:
> On 03/22/2011 06:32 AM, Francesco Pietra wrote:
>> Hi Peter:
>>
>> On Tue, Mar 22, 2011 at 1:42 AM, Peter Freddolino <petefred_at_ks.uiuc.edu> wrote:
>>> Hi Francesco,
>>> Why are you surprised that a *gaff* topology doesn't match *charmm* atom
>>> names and types?
>> I showed that while asking to myself whether autopsf should be able to
>> understand such gaff list or a charmm list is a must.
>
> Hi Francesco,
> psfgen (and thus autopsf) is completely agnostic regarding the force
> field that you use, so long as you supply properly formatted topology files.
>
> In your case, the atom names in your pdb do not match those in the
> topology file. Thus, psfgen has no way of matching atoms in the pdb with
> atoms in the topology, and makes extremely poor guesses instead. You
> need to build your molecule in a way that gives proper atom names.
>
> One way to guarantee proper naming is to build your molecule in
> molefacture, and then use its typing functions to apply gaff
> types/charges and write a topology file. It will handle the naming
> automatically. This requires a little extra work to interface with
> antechamber, as detailed on the molefacture web page. You can also just
> do the renaming manually.

Hi Peter:
For the organic molecules of interest I only have the coordinates and
these should not be changed when building psf/pdb because the
molecules make part of a non-covalent complex with the protein. My aim
is to parameterize these organic molecules to carry out MD with the
complex in a membrane. I have no problems for the protein in a
membrane with namd once the organic ligand is taken off.

As "manually renaming" is appealing for the case in point, what do you
mean exactly? Renaming the starting RIB.pdb at position 13-14 with c3,
os oh ho (i.e., the atom names in the topology RIB.rtf), or their
uppercase equivalent, is not what works. Should I rename both RIB.pdb
and the topology RIB.rtf with atom names agreed by CHARMM ff? If so,
either I find a recent list of atom names (I have one about which I am
uncertain, a huge number of cases) or I should anyway draw the
molecules with molefacture just to get the names and then carry out
the manual renaming in the actual pdb. It seems that paratool would be
a quicker approach (with more reliable partial charges).

thanks
francesco