[gmx-users] Re:gmxdemo: peptide to real protein and boom

[David Mathog]

> 
> Though I see some sense in having the proposed hack, it should be used
with
> care since in most cases the additional groups will be ions or small
subsets
> of atoms, which are better not to be coupled to a heat bath
independently. I
> would advocate not to rely on such an option, but rather check the pdb for
> non-standard groups or use the protein/non-protein groups.

I'm not at all familiar with the purpose of this part of the program
but it struck me as pretty odd that ref_t HAD to be specified for
each group when in each case it was going to be the same temperature.
Also it wasn't at all clear to me what the groups were.  The log file
certainly didn't include any message to the effect of "here are your
N groups which are...".  That info must be in some file though, where?

The message in the log file immediately preceding the section of
interest says:

Making dummy/rest group for T coupling containing 1 elements.

and there were 3 groups, 2 of which were presumably protein and sol,
so maybe that dummy/rest group is the mysterious 3rd group?  Note
that the dummy/rest message doesn't give any kind of hint as to what
that 1 element might be.  The demo script generates this message on
a run of grompp and the pr.mdp file it read doesn't specify a 3rd group
in any way.

Which is a long winded way of saying that grompp creates
a mysterious group, gives no indication of what that group is, what
is in the group, or what the group is called, and then expects
ref_t/tau_t values from the input file!  My work around was to
supply ref_t/tau_t values automatically, but I agree it would be
better if grompp didn't make the extra group(s) in the first place,
or if it did, at least told you enough to fix the problem.

Regards,

David Mathog
mathog at caltech.edu
Manager, Sequence Analysis Facility, Biology Division, Caltech