Hbond wrong number possible bonds

GROMACS version: 2020.4
GROMACS modification: No
I simulated a mixture of diethylene glycol, triethylene glycol, and tip4p2005 water with OPLSAA. I ran gmx hbond to find the intramolecular hydrogen bonding within diethylene glycol. From an index file I made there are two groups from the same molecule: one is the hydroxy oxygens and hydrogens, the other is the ether oxygen. It says there are three possible bonds. I figure there should be two, one from each hydroxy group to the ether oxygen. Any insight is appreciated.

here’s the output:
Specify 2 groups to analyze:
Selected 5: 'PEG_OA_PEG_HA_&DEG&r_1’
Selected 6: 'PEG_OE
Checking for overlap in atoms between PEG_OA_PEG_HA
&DEG&r_1 and PEG_OE&DEG&r_1
Calculating hydrogen bonds between PEG_OA_PEG_HA
&DEG&r_1 (4 atoms) and PEG_OE&DEG&_r_1 (1 atoms)
Found 2 donors and 3 acceptors

Will do grid-search on 8x8x8 grid, rcut=0.34999999
Frame loop parallelized with OpenMP using 8 threads.
Average number of hbonds per timeframe 0.028 out of 3 possible

Diethylene Glycol has two OHs and and one =O, possible combinations,

  • OH1-OH2
  • OH1-(=O)
  • OH2-(=O)

Since the OHs are in the same group shouldn’t it ignore this combination as it is supposed to be looking for hbonding between the two groups not within them? I don’t want it to include the OH-OH bond, is there a way to remove it?

Is there a way to select one group to only be the donor and the other to only be the acceptor?

Possibly different ways you can do that. You may try to calculate HB between OH groups separately and subtract from total HB count.