Marc has been helping me off-list to get the chamber conversion inside
ParmEd working (many thanks to him!), and we stumbled upon a chamber issue
that may be of interest to anybody using CGenFF to create stream files for
custom residues (particularly Sarah who has been a helpful contributor to
this thread).
After we finally got ParmEd working, the chamber-style topology files it
produced started giving different answers than chamber, which is a bit
worrying. Looking into the problem a bit more, it seems that chamber was
truncating atom type names to 4 characters which ends up potentially
conflating a large number of atom types into a single "slot" in the
parameter tables that chamber stores internally. This may result in
chamber doing something like using the CG324--HGA2 parameter for a bond
between atom types CG321 and HGA2.
From what I can tell at this point, ParmEd is doing the right thing. I
believe this only affects systems in which there are atom types longer than
4 characters long. It's easy enough to check chamber -- just use ParmEd to
create a chamber-style topology file and compare a single-point energy
between the two prmtops (inpcrd files should be compatible with both).
Special thanks to Marc for his help.
All the best,
Jason
On Tue, Aug 12, 2014 at 12:20 PM, Marc van der Kamp <
marcvanderkamp.gmail.com
<javascript:_e(%7B%7D,'cvml',
'marcvanderkamp.gmail.com');>> wrote:
> Thanks Jason!
> No crying necessary - it works fine for me now too!
> Best wishes,
> Marc
>
>
> On 12 August 2014 15:16, Jason Swails <jason.swails.gmail.com
> <javascript:_e(%7B%7D,'cvml','jason.swails.gmail.com');>> wrote:
>
> > On Tue, Aug 12, 2014 at 8:43 AM, Marc van der Kamp <
> > marcvanderkamp.gmail.com
> <javascript:_e(%7B%7D,'cvml','marcvanderkamp.gmail.com');>
> > > wrote:
> >
> > > Hi Jason,
> > >
> > > On 12 August 2014 12:52, Jason Swails <jason.swails.gmail.com
> <javascript:_e(%7B%7D,'cvml','jason.swails.gmail.com');>> wrote:
> > >
> > > > On Tue, Aug 12, 2014 at 1:50 AM, Marc van der Kamp <
> > > > marcvanderkamp.gmail.com
> <javascript:_e(%7B%7D,'cvml','marcvanderkamp.gmail.com');>
> > > > > wrote:
> > > >
> > > > > Thanks Jason - No problem for the testing, it's in my benefit too!
> (a
> > > > > workflow with just needing one pass through ParmEd is preferable!).
> > > > > Anyway, I got the following error now:
> > > > >
> > > > > > chamber -top top_all36_prot_ksi.rtf -param
> > > > par_all36_prot_cgenff_ksi.prm
> > > > > -psf dry.rs1_charmm.psf -crd dry.rs1.pdb
> > > > > Creating chamber topology file from PSF dry.rs1_charmm.psf, RTF
> files
> > > > > [top_all36_prot_ksi.rtf], and PAR files
> > [par_all36_prot_cgenff_ksi.prm]
> > > > > Coords from dry.rs1.pdb. Using CMAP. GB Radius set mbondi.
> > > > > Traceback (most recent call last):
> > > > > File "/Users/marcvanderkamp/Documents/work/ParmEd/parmed.py",
> line
> > > 161,
> > > > > in <module>
> > > > > parmed_commands.cmdloop()
> > > > > File
> > > > >
> > > > >
> > > >
> > >
> >
> "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/cmd.py",
> > > > > line 142, in cmdloop
> > > > > stop = self.onecmd(line)
> > > > > File
> > > > >
> > > > >
> > > >
> > >
> >
> "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/cmd.py",
> > > > > line 221, in onecmd
> > > > > return func(arg)
> > > > > File "<string>", line 1, in <lambda>
> > > > > File
> > > > >
> > >
> "/Users/marcvanderkamp/Documents/work/ParmEd/ParmedTools/parmed_cmd.py",
> > > > > line 141, in _normaldo
> > > > > action.execute()
> > > > > File
> > > > >
> > > >
> > >
> >
> "/Users/marcvanderkamp/Documents/work/ParmEd/ParmedTools/ParmedActions.py",
> > > > > line 3563, in execute
> > > > > parm.load_coordinates(coords)
> > > > > File
> > > > >
> > > > >
> > > >
> > >
> >
> "/Users/marcvanderkamp/Documents/work/ParmEd/chemistry/amber/_amberparm.py",
> > > > > line 940, in load_coordinates
> > > > > for i, atom in enumerate(self.atom_list):
> > > > > AttributeError: 'ChamberParm' object has no attribute 'atom_list'
> > > > >
> > > >
> > > > I'm at a complete loss here. It works fine for me and I can't
> > possibly
> > > > see how you are getting these errors (and I'm not).
> > > >
> > > > One thing that's a little concerning is that you've installed ParmEd,
> > yet
> > > > the modules that it is using still seem to come from your source code
> > > > folder and not from
> > > > /home/marcvanderkamp/Library/Python/2.7/lib/python/site-packages...
> > > >
> > > > Yes, I am almost certain this is where the problems/errors stem from.
> > > I installed parmed as follows:
> > > $ pwd
> > > /Users/marcvanderkamp/Documents/work/
> > > $ rm -rf ParmEd
> > > $ git clone https://github.com/swails/ParmEd.git
> > > $ cd ParmEd
> > > $ python setup.py install --user --install-scripts ~/bin/
> > >
> > > I've now also tried to do the same directly in ~/
> > > This leads to the same errors. So it must be to do with how python
> looks
> > up
> > > the modules? Not sure what I should do to fix this.
> > >
> >
> > OK, I think I solved the problem. When OpenMM is installed on your
> > system, ParmEd loads the OpenMM-enabled prmtop classes by default to
> allow
> > you to leverage OMM's capabilities directly in ParmEd (like computing
> > energies, running simulations, etc.). So on my machine, the ChamberParm
> > class created by chamber was cast to an OpenMMChamberParm class, and that
> > casting provided the required attributes (like atom_list).
> >
> > Since you don't have OpenMM installed, that cast did not occur and you
> got
> > the error you observed (that's the challenge with supporting the
> > flexibility of having numerous packages being optional rather than
> > required). I've fixed this error (which I was able to reproduce when
> > removing OMM from my PYTHONPATH) and it now works fine for me. If it
> > _still_ doesn't work for you I think I'm going to cry.
> >
> > Thanks again,
> > Jason
> >
> > --
> > Jason M. Swails
> > BioMaPS,
> > Rutgers University
> > Postdoctoral Researcher
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org <javascript:_e(%7B%7D,'cvml','AMBER.ambermd.org');>
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org <javascript:_e(%7B%7D,'cvml','AMBER.ambermd.org');>
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Aug 12 2014 - 14:00:02 PDT
