From: Francesco Pietra (
Date: Sun Jun 10 2012 - 12:45:22 CDT

Hi Peter:

>Hi Francesco,
>I should add/elaborate that while the use of a topology file containing
>gaff types is not the source of your current technical problems (which
>is what I'm focusing on now), it will be an absolute disaster if you
>actually simulate a system mixing gaff ligand types with charmm
>protein/lipid types. It will run, but there's no earthly reason why you
>should trust the results. You'd be much better off either using amber
>protein/lipid as well or switching to the charmm general force field for
>your ligand.

>>> On Mon, Mar 21, 2011 at 11:48 AM, Peter Freddolino <> wrote:
>>>> Hi Francesco,
>>>> The key question is, are the CHARMMGEN .inp/.rtf files correct? Either
>>>> they are wrong, or one of the assumptions that autopsf makes is invalid
>>>> for that structure and you need to change your settings. The best way to
>>>> check is to (a) manually inspect the .inp/.rtf files and make sure that
>>>> the atoms and connectivity match your expectations, and (b) try
>>>> generating the ibuprofen using psfgen instead of autopsf so that you
>>>> have finer control over what is being done to the molecule.

Thanks for this advice of last year, to which I have tried to adhere
in the meantime.

However, dealing with organic, unusual molecules (typically "natural
products") is quite often difficult with CHARMM ff. A case in point
is my attempt today to have a guess start from CGenFF (ParamChem) for
a molecule containing a four-membered ring (not very unusual with
natural products). The server skipped the molecule, as four-membered
rings are not recognized.

What I hoped to have is an input for the FFTK VMD toolkit, which can't
be used without a psf file. Again, I have to turn to
antechamber/charmmgen to have such an input for FFTK, should such a
procedure work at all. Otherwise, how could I invent a start input by
analogy when the CGenFF finds none?

I know that I could simply run NAMD with ff12SB/GAFF but certain
situations are best treated by CHARMM ff, and this is present case for
reasons not given above.

francesco pietra