[gmx-users] g_sas and g_hbond

[Marc Vogt]

Using both gromacs 3.2.1 and 3.1.4 I have tried to generate a
plot of the solvent accessible surface area for a sidechain of
a single residue.  The plots indicate the sas for a Trp sidehain
is 2.5 nm^2 +/- 0.3 nm^2 constantly over the entire trajectory.
The result applies for all Trp sidechains observed.

However:
The protein clearly unfolds over the trajectory by visual inspection.
Web based methods for SAS calculation of selected snapshot structures
show variation.  So the real value of the sidechain SAS should not
be constant.  

I also see the same behavior for g_hbond on the whole protein, 
where I choose Protein-Protein
hbonds from the index and I see little change (other than fluctuation
about an average), despite the fact that visual inspection shows and 
do_dssp indicates that significant hydrogen bonding is lost.

I searched the archives and found one similar problem that didn't
have a resolution.  That person was using two chains I believe.
My system is a single chain in solvent.  It seems that with such
drastic conformational change one would see more
variation in such calculations. 

To be more specific as to the methods I used, here are my steps.
1)  make_ndx -f protein.gro -o trps.ndx
    
    I chose "8 & r7" to get the sidechain for residue 7.  Then quit

2)  g_sas -f whole.trr -s test.tpr -n trps.ndx -o area7.xvg
 
    I chose the field created from the above command for "8 & r7"


The same problem occurs using an xtc file instead of trr, for multiple residues,
for different versions of GROMACS (3.2.1 and 3.1.4) on different trajectories.

Thanks for any help,

Marc