[gmx-developers] Bugs in g_sdf

[Mark Abraham]

Ondrej Marsalek wrote:
> Dear GROMACS developers,
> 
> I have recently started using the g_sdf (so far only in mode 1) and I
> like it, but I have noticed two issues with it:
> 
> 1) The refmol structure seems to be wrong if the reference groups are
> longer than one. It does work with one reference molecule only. This
> has been reported before:
> 
> http://lists.gromacs.org/pipermail/gmx-users/2009-March/040364.html
> 
> 2) The normalization of the SDF is wrong. It seems that (at least) the
> number of frames is not accounted for properly. The absolute values of
> the SDF grow with the number of frames used. Also, a brief look at the
> code in src/tools/gmx_sdf.c suggests that the number of frames used is
> not considered at all (all normalization is done with the variable
> normfac).
> 
> I would very much like to see these issues resolved. I'd be glad to
> help, but am totally new to GROMACS on the coding side, so I can't
> provide a patch, at least not now.

Please submit a Bugzilla here http://bugzilla.gromacs.org. The more 
information you provide about your system, computational environment, 
command line, etc. the better. Also, make sure you attach a tarball of 
your inputs and outputs that will replicate your problem - but obviously 
a minimal-length input trajectory, and a .tpr that doesn't have to run 
for a week to observe a problem would both be desirable.

The people who might fix bugs you've found are busy volunteers - you 
need to make it easy for them to see you might have a real problem and 
for them to be able reproduce it. Even if they can't solve it 
immediately, with Bugzilla it'll at least be on the to-do list.

Mark