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

[David Mathog]

> 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! 

ok, found the relevant code.  It's in do_numbering() in readir.c.
It checks there that all atoms have been assigned to an existing group
and if it finds any that aren't it creates a group called "rest"
(apparently, from the comments) and puts the extra atoms in there.
Checking... Yes, the group is called "rest", that name is defined
in do_index() in the same file.  I'd bet though that if
you specify a "rest" group in the parameter group and there are no
atoms in it that will trigger some other failure mode.

So proposed change 1, modify the warning to say the name of the group
explicitly - end users shouldn't have to read code to see what a 
program is doing.

And proposed change 2, emit the names of these atoms so the
end user has a clue where to look for the problem.

And proposed change 3, which eliminates the need for 1 and 2 for files
imported from pdb: if grompp is going to barf on atoms
from a pdb2gmx output file then pdb2gmx should know that and
either warn about these atoms at that point, or optionally just remove
them and tell the end user that it has done so.

Regards,

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