[gmx-users] Amber 14SB force-field problem with residue HID

[Kevin]

Thank you Justin for your reply. I totally understand the risk of using a
user-contributed force-field. And that is why I am trying to fix it (and
test it). Then I rather put my question more as a force-field building
question, asking how pdb2gmx generates the list of impropers from the
topology. Hope you could share your experience on this so I could fix it by
myself.

Thanks,
Kevin

On Thu, 12 Sep 2019 at 19:08, Kevin <mcfc1301 at gmail.com> wrote:

> Dear Users,
>
> I have encountered a problem with the Amber force-field downloaded from
> http://www.gromacs.org/Downloads/User_contributions/Force_fields.
> Whenever I feed this forcefield to pdb2gmx and get a residue HID for
> protein topology, it will cause error message: "No default Proper Dih.
> types". This error can simply be solved by swapping the second and third
> indices on the line number given with the error message.
>
> I tried to dig into the reason behind this error. First, the error line
> contains atomtypes in a sequence of "CT  CC  CV  NA" which CANNOT be found
> in ffbonded.itp. Instead, "CT  CV  CC  NA" can be found in ffbonded.itp and
> I guess that is why my solution worked. However, when I look at the
> topology of HID, I was confused:
>
> [ impropers ]
>     -C    CA     N     H
>     CA    +N     C     O
>     CG   CE1   ND1   HD1
>     CG   NE2   CD2   HD2
>    ND1   NE2   CE1   HE1
>    ND1   CD2    CG    CB
>
> Converting from atom name to atomtypes, there is no either "CT  CC  CV
> NA" or "CT  CV  CC  NA" in impropers. However, "ND1   CD2    CG    CB" is
> very likely to be the folk just written in a reversed sequence. More
> interestingly, I gave a try on the Amber ff99*-ILDN forcefield, the HID
> topology appears to be nearly the same but no error message was produced.
> The error was always found in the second [dihedrals] block, so I assume it
> is referred to impropers. That gave me two questions:
> 1. How does impropers be constructed by pdb2gmx?
> 2. Why different forcefields seem to give different sequence of atomtypes
> for imporpers.
>
> In fact, Q2 makes absolutely no sense, so I think I must be missing
> something.
>
> Any comment is greatly appreciated.
>
> Regards,
> Kevin
> The Ohio State University
>