AW: AW: problem with pdb psf files!

From: Norman Geist (
Date: Tue Sep 04 2012 - 01:59:41 CDT



what is seriously not the problem, is the initial structure as the psf
doesn't care about coordinates.


So are you sure that your psf contains this improper parameters?


I still don't understand why you don't use your program to build up only the
initial structure with your molecules as pdb, and just throw it through
psfgen. Why do you need to create the psf yourself?


As you seem to be familiar with codes, you could check the namd code that
produces the error, maybe add some debug lines of output, and check what
namd is reading from your psf. Could you send us a diff output of a psfgen
generated psf and your own one?


Norman Geist.


Von: [] Im Auftrag
von Jacqueline Schmidt
Gesendet: Montag, 3. September 2012 18:01
An: Norman Geist
Cc: namd
Betreff: Re: AW: namd-l: problem with pdb psf files!



Dear Norman,

Thanks for your reply. Some of these systems are peptides on the clay
surfaces that are bound through periodic box in xy directions. I had to
create the pdb psf files for the surfaces through my own code and then
added another section to put the peptide on it.

The question I asked below is for a peptide in solution (as a stating
point). Vmd loads my pdb psf file with no flaw. All connectivities are
correct. However namd gives me an error looking for an improper coefficient.
What is so wired is that the CT1 C O NH1 is always in anypeptide and I
really wonder why I get this error while the psfgen generated files wouldn't
cause this error. Could this be because my structure is for from the
equilibrium? or it is due to the formatting and I have to rearrange my code
to change the writing sequences as in psfgen? thanks again



From: Norman Geist <>
To: 'Jacqueline Schmidt' <>
Cc: Namd Mailing List <>
Sent: Monday, September 3, 2012 3:18 AM
Subject: AW: namd-l: problem with pdb psf files!



also as a 1st friendly comment, what exactly is the problem when using


Are you able to load the pdb/psf pair in vmd? Maybe vmd will tell you


Usually md tools must somehow identifiy parts of a structure by some fixed
properties. I remember that I got some serious problems when I wrote a
program that had to connect molecules of to pdbs at specific binding sites
by automatic adjusting and fitting. The problems was about amino acids, that
had another order than the default is. This caused that I could not identify
N or C-terminals very comfortable to set the right TER lines there.


And as I know, NAMD also threats some things about hydrogens hard coded. So
maybe you also change something that makes it impossible for namd to
interpret the files correctly. But the developers can maybe tell you more.


You could check the code of namd, to see what maybe happens, or just stay
with the order namd seems to expect.


Norman Geist.


Von: [] Im Auftrag
von Jacqueline Schmidt
Gesendet: Montag, 3. September 2012 04:45
An: namd
Betreff: namd-l: problem with pdb psf files!


Dear Namd users,

I have got a problem with my pdb/psf file running with NAMD. I have done
some coding to transfer the car/mdf (insight format) to the pdb/psf file (
just a friendly first comment: please don't refer me to psfgen because I
have a surface in my system which can not be dealt with top file)


When I load the files into the vmd everything seems to be perfect. However
when I run the files in NAMD I get the error that : CAN'T FIND IMPROPER

the error message is self explanatory; however, I don't understand why I am
getting this error at all, because the indicated atom types are the backbone
atoms which every FF parameters should be already in the parm file.


The only difference between my pdb psf files is that I write the
non-hydrogen atoms in any residue first followed by the list of hydrogen
names/ff-types. Is this a problem??


As a test I made the pdb/psf files with psfgen too. Everything between the
files is again the same except two things 1) the order of the atoms 2) It
does not include the right c-terminal atoms with their FF-type indicated in
the top file. I simply treats the last carbon as C C (name and FF-type)
instead of C CC and only one oxygen at the end which is O O instead of OT1
OC, and the OT2 is not in the file at all (it was in my original pdb that
feeded into the psfgen though).


I hope someone can help me please, I am really confused



This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:22:30 CST