[gmx-users] Umbrella sampling along PCA eigenvector using make_edi

[Hendrik Goeddeke]

Hi everyone,

I want to do umbrella sampling along a PCA eigenvector using 'gmx
make_edi'. The eigenvectors of the protein backbone atoms were computed
using 'gmx covar'. The command for generating the 'sam.edi' file looks
like the following:

gmx make_edi -s topol.tpr -f eigenvec.trr -eig eigenval.xvg -restrain
yes -harmonic yes -ori window1_backbone.gro -flood 1 -tau 0 -Eflnull 1000

The file 'window1_backbone.gro' just contains the backbone atoms of a
snapshot called 'window1.gro', which was used for generating
'topol.tpr'. I changed the eigenvalue of the first eigenvector in
'eigenval.xvg' to 1 to get a force constant of 1000 kJ/(mol*nm^2). When
I then start 'gmx mdrun -ei sam.edi' the output states that the initial
RMSD from the reference structure is not zero:

...
ED: Flooding 1 eigenvector.
ED: Initial RMSD from reference after fit = 0.371082 nm
...

But when I compute the RMSD to the reference structure ('topol.tpr')
manually by using 'gmx rms', the RMSD is zero as expected. This excludes
a possibly wrong PBC representaion of the ED group (backbone atoms). I
looked at the code ('edsam.c') but it seems that the RMSD is also
computed from 'topol.tpr', which was specified in 'gmx make_edi'. I
repeated the procedure many times but could not find any mistake.

Any help is highly appreciated.

Thank you very much in advance!

All the best,
Hendrik