[gmx-users] about average structures

[nicolis at guideo.fr]

Hello,

i have a little problem as I am trying to compute an average of the 
coordination polyhedron around a calcium cation.

I proceed with a md run with Ca2+ solvated in water (G43a1 forcefield).

I then use g_dist with the option -dist to see which water molecules are 
around the calcium and then create index groups for each one and use 
again g_dist without the -dist option to have the variation over time for 
each calcium/water.

I choose a time range where no exchange takes place and create an 
index group including these water oxygens and the calcium cation (9 
atoms in total).

Finally, I run g_covar using this last created index group to have the 
average structure over this time period.

The problem is that when I charge this average.pdb in rasmol and check 
the distances they are not the same as the average over time of the 
distances output from the g_dist (I speak for each distance separately).
Is this a nonlinear effect of the way g_covar computes averaged structure 
or there is something wrong in the way I use it?

Thanks for any suggestion,
Yiannis