Double counter in Water bridge analysis

[xiki-tempula]

Expected behavior

User Irtaza Hassan emailed me providing a case where

        Acceptor(A1)···H(H2)−O(O3)···H(H7)-Donor(D8) 
                             |
                             H(H4)···O(O5)-H(H6)

This should give a first-order water bridge.

Actual behavior

It recognised the following route
A1···H2-O3
O3-H4···O5
O5···H4-O3
O3···H7-D8
Will be recognised as a third-order water bridge

Code to reproduce the behavior

from MDAnalysis.analysis.hydrogenbonds import WaterBridgeAnalysis
import MDAnalysis
from io import StringIO
grofile = '''Test gro file
    7
    1LEU      O    1   1.876   0.810   1.354
  117SOL    HW1    2   1.853   0.831   1.162
  117SOL     OW    3   1.877   0.890   1.081
  117SOL    HW2    4   1.908   0.828   1.007
  135SOL     OW    5   1.924   0.713   0.845
    1LEU      H    6   1.997   0.991   1.194
    1LEU      N    7   2.041   1.030   1.274
   2.22092   2.22092   2.22092'''
u = MDAnalysis.Universe(StringIO(grofile), format='gro')
wb = WaterBridgeAnalysis(u, 'resname LEU and name O', 'resname LEU and name N H', order=4)
wb.run()
print(wb.timeseries)

[[(0, 1, ('LEU', 1, 'O'), ('SOL', 117, 'HW1'), 1.9450972623464, 144.11627508942746), 
  (2, 5, ('SOL', 117, 'OW'), ('LEU', 1, 'H'), 1.933132206990047, 162.05564720359993), 
  (3, 4, ('SOL', 117, 'HW2'), ('SOL', 135, 'OW'), 1.9931120422844908, 166.0596405239858), 
  (4, 3, ('SOL', 135, 'OW'), ('SOL', 117, 'HW2'), 1.9931120422844908, 166.0596405239858), 
  (2, 5, ('SOL', 117, 'OW'), ('LEU', 1, 'H'), 1.933132206990047, 162.05564720359993)]]

Current version of MDAnalysis

• Which version are you using? I will fix this for both master and develop branch