Hi,
I am also no expert on Gromacs topologies, but it seems like the [ cmap
] entry only defines which atoms should have CMAP terms at all, while
the [ cmaptypes ] entry defines a mapping from atom types to CMAP type.
For example, one definition in my .top file looks like this:
[ cmaptypes ]
C N XC C N 1 24 24\
(followed by 576 numbers)
To adapt my example topology, I copied the XC atom type to generate XC1
and XC2, and added the corresponding entries. I did not need any further
changes.
Best,
Franz
On 21.06.23 16:52, Carlos Simmerling wrote:
> Hi Franz,
> thanks that's very helpful. Amber uses the residue name inside Leap
> while it is building the topology, but inside the final topology file
> itself the cmaps are specified using atom number and cmap id (pointing
> to a table of cmaps that are stored in the topology file itself), so
> no residue name is needed. I'm not sure why Jason suggested that the
> prmtop uses the residue name (unless I misunderstood his comment on
> the github issue). It sounds like converting what is in the Amber
> prmtop to something needed for gromacs should be straightforward. As I
> said I'm not familiar myself with how gromacs implements cmaps or how
> they are stored. Does it assign cmaps based on atom type?
> carlos
>
> On Wed, Jun 21, 2023 at 10:29 AM Franz Waibl via AMBER
> <amber.ambermd.org> wrote:
>
> Dear Carlos, dear Amber users,
>
> thanks for the reply. I am aware that Amber is free for academic use
> (thanks for that, by the way!), but it would still be useful to be
> able
> to run ff19SB simulations on Gromacs, e.g., to use algorithms that
> are
> not implemented in Amber.
>
> Gromacs does support multiple CMAPs. If I understand correctly, the
> problem is that ff19SB CMAP terms depend on the residue name in
> addition
> to the atom type. To convert the topology correctly, one would
> have to
> define separate atom types for the CA atom of each amino acid
> (XC1, XC2,
> ...). This also seems to be what CHARMM-GUI is doing (see the Github
> issue mentioned before).
> In fact, I ran a very short test simulation in Amber and compared the
> results to energy values from gmx mdrun -rerun. After manually
> tweaking
> the topology as described above, I was able to obtain consistent CMAP
> energies between Amber and Gromacs.
>
> However, it is very tedious to do this manually for a larger
> system, so
> a ParmEd-based solution would be highly appreciated.
>
> Best regards,
> Franz
>
>
> On 21.06.23 15:14, Carlos Simmerling wrote:
> > My understanding is that the issue is on the gromacs end, not
> parmed.
> > I'm not a gromacs user but you could ask them if multiple cmaps are
> > supported. If yes, then we can work to update the conversion
> process.
> > Alternatively, Amber is free and has a good library of tutorials.
> > Carlos
> >
> > On Wed, Jun 21, 2023, 8:59 AM Franz Waibl via AMBER
> > <amber.ambermd.org> wrote:
> >
> > Dear Amber users,
> >
> > I would like to run simulations with the Amber-ff19SB
> forcefield on
> > Gromacs. I have seen an Issue on Github
> > (https://github.com/ParmEd/ParmEd/issues/1292) stating that
> ParmEd
> > does
> > not handle the CMAP term conversion correctly. I also ran
> some tests
> > with the ParmEd version in Ambertools23 installed via conda,
> and it
> > seems that the same problem occurs: Only one set of CMAP
> > parameters is
> > written to the Gromacs topology, rather than different
> parameters for
> > each amino acid.
> >
> > Do you know if this problem can be avoided somehow in
> ParmEd? If no,
> > what is the recommended way to obtain ff19SB topologies for
> Gromacs?
> >
> > Thanks for your help.
> >
> > Best regards,
> > Franz
> >
> > --
> > ETH Zürich
> > Franz Waibl
> > Lab. für Physikalische Chemie
> > HCI G 227
> > Vladimir-Prelog-Weg 1-5/10
> > 8093 Zürich
> >
> > Telefon +41 44 632 55 04
> > franz.waibl.phys.chem.ethz.ch
> >
> >
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
>
> --
> ETH Zürich
> Franz Waibl
> Lab. für Physikalische Chemie
> HCI G 227
> Vladimir-Prelog-Weg 1-5/10
> 8093 Zürich
>
> Telefon +41 44 632 55 04
> franz.waibl.phys.chem.ethz.ch
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
ETH Zürich
Franz Waibl
Lab. für Physikalische Chemie
HCI G 227
Vladimir-Prelog-Weg 1-5/10
8093 Zürich
Telefon +41 44 632 55 04
franz.waibl.phys.chem.ethz.ch
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Jun 21 2023 - 08:30:02 PDT
