[gmx-users] gmx rms vs. trjconv

[mhuhtala at abo.fi]

I ran into what seems to be a bug, namely that the -skip option in gmx  
rms doesn't seem to do anything. The output always has the RMSD values  
for all frames, even if -skip is specified. I filed a bug  
(https://redmine.gromacs.org/issues/2565).

Then I resorted to trjconv to do the skipping before using gmx rms,  
and figured I might as well do the fitting of the structures at that  
stage, since it has the added benefit of calculating the fit just  
once. I need RMSD values for different parts of a protein, but using  
the same fit to the reference structure. To my surprise, the RMSD  
values are different when the frames are fitted with trjconv and then  
calculated with gmx rms -fit none, as compared to just using gmx rms  
with -fit rot+trans. In my case, when using the protein backbone for  
both the fit and the RMSD calculation, the RMSD values are roughly 0.5  
Å larger for the trjconv-fitted trajectory, but it's not an exact  
systematic shift.

Am I missing something obvious here? Are these two procedures supposed  
to do the same thing or not?

gmx rms alone:

    gmx rms -s topol.tpr -f traj_comp.xtc -fit rot+trans -skip 100 -n index.ndx

    (-skip does not work; select protein backbone for fit and for RMSD)

trjconv + gmx rms:

    gmx trjconv -f traj_comp.xtc -s topol.tpr -fit rot+trans -skip 100  
-n index.ndx

    (select protein backbone for fit)

    gmx rms -s topol.tpr -f trajout.xtc -fit none -n index.ndx

    (select protein backbone for RMSD calculation)