From: David A. Horita (
Date: Thu May 12 2011 - 16:51:39 CDT

I ran into this problem in 2008 with some modified CHARMM files - 4 characters works for atom names from a VMD perspective, but the newer charmm rules allow up to 8 characters. The CG2R51 is causing the problem. It's the A8, not the G14.6.

John might be able to dig out some communication with me and Alex MacKerell about this from 2008-

CHARMM is currently being modified to accept 8 characters, which will be the standard into the future. That's why I've gotten in touch with you to insure that NAMD/VMD can handle 8 characters. Youngdo Won and Bernie Brooks are the ones doing the modification. I'm ccing this to them so they can comment if necessary.

thanks, alex

John Stone wrote:
> Hi,
> I think the root of the problem here is that the CHARMM PSF format
> uses integer indices for the atom type field, and psfgen and NAMD use
> the X-PLOR scheme of using that field as an atom type string.
> The CHARMM PSF formats only allow 4 characters for the atom type,
> which is fine if you use an integer index as they use in their Fortran
> format string, but that's insufficient for the actual atom type
> string. The CHARMM EXTended PSF format is (if CMAP,CHEQ are also included):
> '(I10,1X,A8,1X,A8,1X,A8,1X,A8,1X,I4,1X,2G14.6,I8,2G14.6)'
> The PSF file you sent illustrates this problem clearly.
> Note that the columns are shifting, which should never occur given
> that all existing variants of the PSF format uses fixed field widths:
> 24 !NATOM
> 1 O1 124 INI1 C1 CC3161 0.140000 12.0110 0
> 2 O1 124 INI1 H1 HCA1 0.090000 1.0080 0
> 3 O1 124 INI1 O1 OC311 -0.650000 15.9994 0
> 4 O1 124 INI1 HO1 HCP1 0.420000 1.0080 0
> 5 O1 124 INI1 C5 CC3161 0.140000 12.0110 0
> 6 O1 124 INI1 H5 HCA1 0.090000 1.0080 0
> There's no existing PSF variant that I'm aware of that will accept
> more than 4 characters for the atom type info, so we will either need
> to create a NAMD-specific variation that eliminates these fixed field
> width issues entirely, or adopt a new file format instead of
> continuing to (ab)use the PSF format.


David A. Horita, Ph.D.
Department of Biochemistry
Wake Forest University School of Medicine
Winston-Salem, NC 27157-1016
Tel: 336 713-4194
Fax: 336 716-7671

From: [] On Behalf Of Grant Schauer
Sent: Thursday, May 12, 2011 3:16 PM
To: John Stone; Axel Kohlmeyer; Grant Schauer;
Subject: Re: vmd-l: Large positive non-integer charge

Hi John,

Yes, that must be it. Large atom names from CGenFF appear to be pushing charges over, e.g.

35 O1 4 RES CL10 CS -0.040000 12.0110 0
36 O1 4 RES CL11 CG2R51 -0.160000 12.0110 0

It would be easy enough to correct by hand, if I can fit the atom name in the column. Do you happen to know the required column widths or could you point me to the PSF documention? From this post

it looks like "G14.6" but I don't know how to interpret that...

Thanks for your help

On Thu, May 12, 2011 at 3:00 PM, John Stone <<>> wrote:
 I suspect it is likely that the PSF might be misformatted, and
that one of the values is going outside of the fixed-width fields.
Please send me a gzipped copy of the PSF file and I'll gladly have a look.

 John Stone<>

On Thu, May 12, 2011 at 02:56:44PM -0400, Axel Kohlmeyer wrote:
> On Thu, May 12, 2011 at 2:16 PM, Grant Schauer <<>> wrote:
> > Hello all,
> >
> > I'm trying to solvate a small system consisting of a small peptide patched
> > to a novel residue built from CGenFF definitions. When I attempt to
> > autoionize the system after solvating, I get a warning that I have a
> > non-integer charge. In fact, it's measuring +67.08000009506941e.
> >
> > I get the same number with the following commands (regardless of whether
> > it's solvated):
> >
> > set sel [atomselect top all]
> > set charge [vecsum [$sel get charge]]
> >
> > But when I add up the charges from the PSF by hand, I get exactly the
> > integer (+2) that I expect, as specified in my topology file. Autopsf runs
> > smoothly, with no complaints that I'm seeing. I'm running VMD 1.9.
> >
> > Any suggestions? Thanks,
> VMD uses single precision floating point numbers,
> which have only about 7 valid digits in the mantissa.
> for a system with a large enough number of charged
> atoms, this kind of non-integer rest may be the consequence
> of simple rounding and truncation errors.
> axel.
> > Grant
> >
> >
> --
> Dr. Axel Kohlmeyer
> Institute for Computational Molecular Science
> Temple University, Philadelphia PA, USA.

NIH Resource for Macromolecular Modeling and Bioinformatics
Beckman Institute for Advanced Science and Technology
University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801>           Phone: 217-244-3349<tel:217-244-3349>       Fax: 217-244-6078<tel:217-244-6078>