Re: FW: psf generation for a small non-peptide molecule - again

From: Peter Freddolino (
Date: Mon Sep 29 2008 - 19:43:21 CDT

Hi Vlad,
it sounds like there are two separate issues here.

The first is how to get autopsf/psfgen to properly work on your
molecule. What is the exact error that makes you say you get atoms
appearing as unknowns? What topology file are you using as input to
autopsf? Did you generate it with molefacture? If you have all of the
atom names, types, charges, and connectivities specified in molefacture,
you can export a topology file from the File menu and use it as input to
autopsf, which should fix some set of problems.

The second problem here is how to get proper parameters for your
molecule. The charmm forcefield is not designed to have generic types
such as "carbonyl oxygen". The way to properly parameterize a molecule
for use with charmm is to go through the process described in the charmm
forcefield papers, involving ab initio calculations for the bonded terms
force constants and equilibrium values, and fitting the atomic charges
to reproduce some QM data. VMD includes the paratool plugin
( to assist in
this process, as it is certainly nontrivial. To get a feel for the kind
of issues involved here you may also want to have a look at the
Parameterizing a novel residue plugin
although the procedure described here will not yield parameters one can
comfortably use with the charmm forcefield unless you're parameterizing
something that is a minor perturbation on an existing molecule. Usually
for a case like yours one ends up dividing the molecule into a set of
smaller model compounds which are then parameterized using the charmm

Hope this is helpful...


CHINDEA Vlad wrote:
> Hello
> Sorry for bothering you with such trivial questions, but I would
> really appreciate some ideas regarding these issues. The PDB that I am
> working with was exported from crystallographic CIF file, and atom
> type allocation was done by hand. Especially I am not sure for the
> proper type of the oxygen in carbonyl groups (I used type no. 70 'O')
> and nitrogene in tertiary amine groups (I have used type 'NR1'). I
> have also corected the file in Molefacture as atom type 'OS' from the
> PDB was converted to element 'OS' and exported an XBGF file which I
> used further for AutoPSF, but still 10 atoms apear us unknowns
> (unparametrized) Please find attached the PDB and the selection of the
> atoms with problems. Can you please give me some ideas about what
> could be wrong with those atom types allocations ? Also while runing
> Autopsf I got the message "I couldn't find the original pdb to assign
> connectivity. Using autogenerated connectivity only" although I had
> both the XBGF and the PDB file loaded in VMD.
> Many thanks
> > From:
> > To:
> > Subject: psf generation for a small non-peptide molecule
> > Date: Fri, 26 Sep 2008 00:18:40 +0300
> >
> >
> > Hello everybody
> >
> > As I am new to the NAMD program I have run into problems with psf
> generation.
> >
> > As top_all27_prot_lipid topology file supplied with NAMD tutorials
> is specificaly made for proteins and lipids I am not sure I am using
> correct atom types for my molecule.
> > Chemically-wise this molecules contains Sp3 carbons, polar and
> nonpolar hydrogens, hydroxyl, ester and ether oxygens, tertiary amine
> nitrogens. Particullary I am not sure what are the proper atom types
> for ether oxygen and tertiary amine nitrogen in this topology file. Do
> you have any suggestions ?
> >
> > Further more is there any other, more general maybe, topology file
> that should be more suitable for this kind of modelling tasks ?
> >
> > Many thanks and kind regards
> >
> > Vlad Chindea
> > _________________________________________________________________
> > Connect to the next generation of MSN Messenger
> >
> ------------------------------------------------------------------------
> Connect to the next generation of MSN Messenger Get it now!
> <>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:49:54 CST