[gmx-users] RE: Bonds and atom-pairs in g_hbond

[Andrew DeYoung]

Hi again,

I just realized that the output from g_hbond using the -num flag (by
default, the file name is hbnum.xvg) also seems to make a distinction
between "hydrogen bonds" and "pairs within the distance."  For example, here
are the first few lines of this output file:

@ s0 legend "Hydrogen bonds"
@ s1 legend "Pairs within 0.227905 nm"
         0          39           1
      0.05          40           2
       0.1          40           2
      0.15          40           1
       0.2          41           0
      0.25          38           3
       0.3          39           2

where 0.227905 nm is the value that I give to -r.  Why are there so many
more "hydrogen bonds" than "pairs within -r", if the latter seems to be a
necessary but not sufficient condition for the former?

Best wishes!

Andrew DeYoung
Carnegie Mellon University

-----Original Message-----
From: Andrew DeYoung 
Sent: Monday, June 11, 2012 5:31 PM
To: gmx-users at gromacs.org
Subject: Bonds and atom-pairs in g_hbond

Hi, 

I am analyzing hydrogen bonds using g_hbond.  I have selected two
non-overlapping groups; one is hydroxyl groups (OH, where O is the donor and
H is the hydrogen), and the other is oxygens on PO4^3- (where F is the
acceptor).  I am using the switch -noda, which tells the program to take the
-r value as the Hydrogen-Acceptor distance (here, H-O), rather than the
Donor-Acceptor distance (here, O-O).  My trajectory is 10 ns long.

In the output after the calculation, I see the following:

Found 233 different hydrogen bonds in trajectory
Found 224 different atom-pairs within hydrogen bonding distance

Can you please help me to understand the terms "hydrogen bond" and
"atom-pair within hydrogen bonding distance" conceptually?

I think that in order for a hydrogen and an acceptor to be considered
hydrogen bonded, two criteria must be satisfied: (1) the distance (in this
case, the distance between the hydrogen and the acceptor, since I am using
-noda) must be less than the cutoff distance -r, and (2) the
Acceptor-Donor-Hydrogen angle must be less than the cutoff angle -a.  So
distance and angle are the two criteria that must be met, according to
g_hbond and most definitions of hydrogen bonding.

So, apparently, g_hbond found 233 hydrogen bonds satisfying both the
distance and angle criteria.

But, g_hbond "found 224 different atom-pairs within hydrogen bonding
distance."  My guess is that this means that there are 224 atom pairs
satisfying the distance criterion (i.e., Hydrogen-Acceptor pairs within the
cutoff distance, -r).  But if my interpretation of the term "atom-pair
within hydrogen bonding distance" is correct, how can there be 9 _fewer_
atom pairs that satisfy only the distance criterion (and not the angle
criterion), than those that satisfy _both_ the distance criterion _and_ the
angle criterion?

Probably, I have the wrong pictures of the terms "hydrogen bond" and
"atom-pair within hydrogen bonding distance" in my head.

Thank you for your time!

Andrew DeYoung
Carnegie Mellon University