[gmx-users] spurious dihedral warnings

[Berk Hess]

>From: mernst at tricity.wsu.edu
>Reply-To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>To: gmx-users at gromacs.org
>Subject: [gmx-users] spurious dihedral warnings
>Date: Thu, 2 Feb 2006 13:55:34 -0800 (PST)
>
>Hello,
>
>I am using gromacs 3.3.
>
>I have a line in my .itp file that looks like this before I actually define 
>a different
>"B" state:
>
>43 amber99_40      2    8oG     N9     43      0.111      14.01    
>amber99_40 0.111 14.01
>
>I can generate a .tpr file with no warnings. However, when I give it a 
>different partial
>charge in the "B" state:
>
>43 amber99_40      2    8oG     N9     43      0.111      14.01    
>amber99_40 0.0577 14.01
>
>Then I get the following warning from grompp:
>
>WARNING 1 [file "trimer_A.itp", line 942]:
>   No default Proper Dih. types for perturbed atoms, using normal values
>
>Line 942 looks like this:
>
>41    43    44    58     1    nucleic_imp_10
>
>The warning persists even when I change it to this:
>
>41    43    44    58     1    nucleic_imp_10 nucleic_imp_10
>
>Using gmxcheck, the only difference I find between the warning-bearing and 
>warning-free
>cases is the different partial charge in the "B"-state on atom 43.
>
>Further, it turns out that I can also trigger the warning by changing the 
>mass of the
>atom in the "B"-state.
>
>In http://www.gromacs.org/pipermail/gmx-users/2006-January/019269.html it 
>seems that
>Jonathan Moore has encountered this problem before, found no fix, and 
>decided to just
>ignore the warnings. I will be ignoring them too for now, but I don't like 
>ignoring
>warnings. Aren't proper dihedrals defined by the atom types involved? How 
>could changing
>the atom's partial charge or mass result in dihedral misrecognition?

These warnings are not spurious.

The nucleic_imp_10 definition only defined parameters
for the A-state topology.
If the atoms involved in the interaction are identical in
the A and B state grompp assumes the bonded interaction
is also identical (unless you explicitly set different B state
parameters).
If one or more of these atoms have any difference in
the A and B state, and you only specify A-state parameters
in the dihedral, I think it is reasonable to issue a warining,
as with different atom properties, the user might have forgotten
to add different dihedral parameters.

Specifying nucleic_imp_10 twice has no effect,
as the the definition ends with a comment, so the second
definition will be hidden in the comment of the first.

That said, it might be useful to have a way to avoid
these warnings, but we also need to make sure the user
will not generate unwanted interaction parameters
without notice.

Berk.