[gmx-users] g_chi non-standard definition of phi and psi

[chris.neale at utoronto.ca]

g_chi with the -phi and -psi options appears to use non-standard  
definitions of those angles. As best as I can discern from my  
measurements with VMD and g_chi,

GROMACS g_chi:
phi: H-N-Ca-C minus 180deg
psi: N-Ca-C-O minus 180deg

the file chi.log lists the atoms that I have outlined above and I am  
adding the "minus 180deg" because of my VMD-g_chi comparisons.  
However, the numbers work out exactly based on the assumption that  
this is what is occurring.

What I thought was standard:
phi: C[i-1]-N-Ca-C
psi: N-Ca-C-N[i+1]

Basing alanine dipeptide dihedral umbrella restraints on my  
understanding of the backbone dihedrals and then using g_chi to  
determine values from the .xtc for WHAM has in this way lead to a PMF  
whose magnitude is too large and that is dependent on the actual force  
constant used. I have not yet calculated the actual values directly,  
but I assume that this is the problem. I noticed this because when i  
plot the dih.rest. energy output from g_energy and scatter plot this  
with the actual dihedral value from g_chi for a particular umbrella I  
get very significant scatter about the expected parabola.

Perhaps I have my definitions backward. Has anybody else run into  
this? I was involved in a couple posts about a month ago regarding  
dihedral pmfs and other people seem to be getting appropriate results.
http://www.gromacs.org/pipermail/gmx-users/2007-January/025555.html

Thank you.
Chris.