I can't comment on whether the numbers you are getting are accurate, but in
principle there is nothing wrong with a negative amplitude term. these are
not "barriers" since they are simply corrections terms to the underlying
model. it is quite reasonable that sometimes the barrier in the underlying
model is too large, and needs to be reduced. it is a common misconception
that the dihedral terms are the rotational energy profile - they are not.
they are just corrections.
On Wed, May 31, 2023 at 8:11 AM Robert Heße via AMBER <amber.ambermd.org>
wrote:
> Dear AMBER user,
>
> recently i wrote to this mailing list regarding the fitting of missing
> atom types for modified nucleotides and after some tries i was able to
> generate a frcmod file, which takes into account both Gaff and
> AMBER(/OL15) atom types. When i wanted to refine these mixed parameters
> with mdgx, i encountered negative barrier terms:
> ----
> Generated by mdgx executing 07_fit.mdgxin.
> MASS
>
> BOND
>
> ANGL
>
> DIHE
> OS-CT-na-cc 1 46.06460 0.0 -2.0 same as X -c3-na-X , penalty
> score= 0.0, Fit by mdgx
> OS-CT-na-cc 1 1.57990 0.0 1.0 same as X -c3-na-X , penalty
> score= 0.0, Fit by mdgx
> OS-CT-na-ca 1 -30.72963 0.0 -2.0 same as X -c3-na-X , penalty
> score= 0.0, Fit by mdgx
> OS-CT-na-ca 1 34.91051 0.0 1.0 same as X -c3-na-X , penalty
> score= 0.0, Fit by mdgx
> CT-CT-na-cc 1 41.14259 0.0 -2.0 same as X -c3-na-X , penalty
> score= 0.0, Fit by mdgx
> CT-CT-na-cc 1 10.25461 0.0 1.0 same as X -c3-na-X , penalty
> score= 0.0, Fit by mdgx
> CT-CT-na-ca 1 -28.74262 0.0 -2.0 same as X -c3-na-X , penalty
> score= 0.0, Fit by mdgx
> CT-CT-na-ca 1 54.87103 0.0 1.0 same as X -c3-na-X , penalty
> score= 0.0, Fit by mdgx
> H2-CT-na-cc 1 39.28179 0.0 -2.0 same as X -c3-na-X , penalty
> score= 0.0, Fit by mdgx
> H2-CT-na-cc 1 5.71388 0.0 1.0 same as X -c3-na-X , penalty
> score= 0.0, Fit by mdgx
> H2-CT-na-ca 1 -32.28837 0.0 -2.0 same as X -c3-na-X , penalty
> score= 0.0, Fit by mdgx
> H2-CT-na-ca 1 49.39032 0.0 1.0 same as X -c3-na-X , penalty
> score= 0.0, Fit by mdgx
>
> IMPROPER
>
> NONB
> ----
> Those dihedral are for C2'/O4'/H1'-C1'-N9-C4/C8 of a modified
> deoxyguanosine nucleotide. I tried to adjust the Tutorial for mdgx for
> this as much as possible. I generated a total of 144 structures (2.5°)
> for sampling and then performed single point calculations via Orca. The
> Orca runs terminated normally and extracting the energies via the Speval
> Module went successful as well. The fitting report from the mdgxout-file
> also looked good, showing an improvement of the parameters:
> (1.) Energies of each system, according to the fitted parameters. Units
> on
> errors are kcal/mol. The original model's energies are compared to
> target energies after applying an offset to make the average of
> each set
> of energies for any given system equal (see Section 3).
>
> RMS Error Correlation Fitted Fitted
> Model
> System Orig Fitted Orig Fitted Terms Energy
> Diff.
> ---------------- ------- ------- ------- ------- ------ -------
> -------
> /home/hesse/DNAz 8.1688 1.5465 0.0562 0.9795 12 0.00
> 8.010
>
> <++>-----------------------------------------------------------------------<++>
> Sorry for the detailed information so far but im a but confused by the
> generation of negative energy barriers as i couldn't find a mistake in
> the files so far.
>
> With best regards,
> Robert Heße
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed May 31 2023 - 05:30:03 PDT
