[gmx-users] CHARMM nonbonded parameters and grompp output
Justin A. Lemkul
jalemkul at vt.edu
Fri Feb 22 01:39:27 CET 2008
Thanks for the detailed and informative reply, Mark (as always). I have decided
there is probably something fundamentally wrong with what I've been doing. I've
been at this for several months, every time I get a few free minutes from my
main work, so I lost track of what I did at the very beginning (oops). I tried
to start from scratch, using your scripts to get ffcharmmbon.itp and
ffcharmmnb.itp, since I know I have tried different iterations of your scripts
and the awk scripts.
When I tried to start from scratch using your convert_charmm_to_gromacs.pl on
the CHARMM .prm file, but it exited with this:
Sin of delta not zero, hope this is the only dihedral of
type X CS SS X
(sin delta = 0.00349065141522373)
The ffcharmmnb.itp file is incomplete, and I suspect ffcharmmbon.itp is as well.
Any idea what's going on?
Thanks, as always.
Quoting Mark Abraham <Mark.Abraham at anu.edu.au>:
> Justin A. Lemkul wrote:
> > Hi all,
> > As a number of others have attempted, I am exploring the use of the CHARMM
> > fields in Gromacs. I have read about a number of difficulties throughout
> > list archive, but I am seeing something that thusfar it seems no one has
> > reported, regarding nonbonded parameters. Let me tell you what I've done
> > far, and perhaps someone can shed some light on what's going on.
> > I downloaded the ffcharmm* files from the old User Contribution site,
> > a link I found in the archive. I created a .hdb file (for ease of use) and
> > edited the .tdb files, as they contained some inconsistencies in atom
> > and formatting. I converted the CHARMM27 force field parameters from Alex
> > MacKerell's site using a script that was also made available through User
> > Contributions, giving me ffcharmm.itp, ffcharmmbon.itp, and ffcharmmnb.itp.
> > far, so good. For the moment, I am also approximating the Urey-Bradley
> > potential by using distance restraints, using scripts that came along with
> > ffcharmm package.
> The two packages are not intended to work together in this way. Both
> have to work around the fact that pdb2gmx will not write U-B potentials
> natively. Yuguang Mu worked around this using distance restraints,
> whereas I'm using the implemented U-B bond type that DvdS mentioned.
> Yuguang's scripts predate mine, and I recall when I implemented the
> same, I had to edit the GROMACS 3.3(?) sources to actually make U-B bond
> types available. Presumably that need motivated his choice of
> work-around. That source editing is no longer needed.
> Anyway, the point is that using any of Yuguang's awk or shell scripts is
> not supported with the ffcharmm*itp files produced by my conversion
> script. I've no idea what the combination would do, and no interest in
> finding out! :-) The rudimentary documentation in the headers of my
> scripts discuss the necessary issues.
> > I was able to produce a topology for hen egg white lysozyme (a decent
> system to
> > test, I thought), with correct disulfides, charges, etc. and life was good.
> > ran grompp to attempt an in vacuo energy minimization, and saw this among
> > output:
> > processing topology...
> > Generated 0 of the 8646 non-bonded parameter combinations
> > The rest of the grompp output indicated no errors or warnings (aside from a
> > charge, which is OK for now). I reasoned that there should be _some_ form
> > nonbonded interactions within the protein, correct?
> Yes, but that doesn't mean they have to be "generated". grompp allows
> pairwise interactions to be listed explicitly in [nonbond_params], and
> otherwise generates them from the combination rule chosen in [defaults]
> and the atom type parameters in [atomtypes]. Various force fields
> implemented in GROMACS use different combinations of the above to get
> their jobs done. As I note in the header to my ffcharmmnb.itp, the
> former is necessary in CHARMM, because they have an extra constant
> factor of 2 in their 6-12 potential that can't be worked around with any
> of the GROMACS generating options. So I generate all possible
> combinations in [nonbond_params] and grompp finds those.
> > I read about 1-4
> > interactions (i.e. OPLSAA generates these by scaling, so they are not
> > explicitly included in ffoplsaanb.itp), but such scaling is reportedly not
> > in the CHARMM force field, and thus it was no surprise to find [
> > ] within ffcharmmnb.itp. This brings me to my question: is grompp not
> > these parameters? And if so, why?
> "finding" != "generating". You can gmxdump your output .tpr file to see
> what nonbonded parameters are being used. Remember numbering will then
> start from zero.
> CHARMM uses some dedicated 1-4 parameters for some atom types. My
> scripts deal with this automatically; I've no idea what Yuguang's
> scripts do here.
> > I noticed that ffcharmmnb.itp is formatted much like the GROMOS force field
> > files (ffgmxnb.itp, ffG*nb.itp) and saw something that is not mentioned in
> > manual. The first line under [ atomtypes ] in each of the GROMOS force
> > has formatting like:
> > ;name at.num mass charge ptype c6 c12
> > O 8 15.99940 0.000 A 0.22617E-02 0.74158E-06
> > The at. num column is not mentioned in the manual (Gromacs version 3.3).
> Is it
> > needed for proper interpretation of the *nb.itp files? Such a column is
> > missing in my ffcharmmnb.itp file. The [ nonbond_params ] and [ pairtypes
> > sections seem to have correct formatting, so I am not sure that this
> > (potential) inconsistency among the [ atomtypes ] is causing the issue.
> There are a number of optional fields in entries to this directive that
> allow work-arounds for different force field issues. See comment in
> push_at() in src/kernel/toppush.c
> > Thanks for reading (yet another) long email from me, and thanks in advance
> > anyone has any ideas as to what's going on.
> No problem. I will eventually get around to better documentation for my
> stuff :-)
> gmx-users mailing list gmx-users at gromacs.org
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
Justin A. Lemkul
Graduate Research Assistant
Department of Biochemistry
jalemkul at vt.edu | (540) 231-9080
More information about the gromacs.org_gmx-users