[gmx-users] g_hbond seems to detect different hbonds in the same trajectory

[Mark Abraham]

Jean-Paul Becker wrote:
> Dear users,
> I computed the hydrogen bonds between solvent and protein in a 200 ps fraction of a MD trajectory
> with g_hbond in two different ways:
> in a single step with the command:
> g_hbond -b 400 -e 600 -hbn -hbm -g 400_600.log -num -n file.ndx -s file.tpr -f file.xtc
> in four steps with the commands:
> g_hbond -b 400 -e 450 -hbn -hbm -g 400_450.log -num -n file.ndx -s file.tpr -f file.xtc
> g_hbond -b 450 -e 500 -hbn -hbm -g 400_450.log -num -n file.ndx -s file.tpr -f file.xtc
> g_hbond -b 500 -e 550 -hbn -hbm -g 400_450.log -num -n file.ndx -s file.tpr -f file.xtc
> g_hbond -b 550 -e 600 -hbn -hbm -g 400_450.log -num -n file.ndx -s file.tpr -f file.xtc
> I made several attempts with different trajectories, different time range and with different number of steps for the second method and i noticed there were often more hydrogen bonds (let's say 20 for a total number of 1000 bounds) described in the files resulting from the several steps method compared to the one step one.
> I was expecting g_hbond to detect the same hydrogen bonds between the same residues on a precise trajectory either i made the computation in one step or using different time frames.
> I noticed this using either gmx 3.3.3 or gmx 4.0.2.
> I would really appreciate any light to clarify these different results with many thanks.

I expect you are double-counting frames at your time range extrema in 
your second method. Actually looking at the output of -num would make 
this clear.

Mark