[gmx-users] Topology for covalent bonds across periodic boundaries

[Rachel Baarda]

Dear Gromacs users,

I am trying to simulate an "infinite" protein chain by covalently bonding
the last residue of my structure to the first residue across the periodic
boundary. I have modified the topology file to include these bonded
interactions and have specified "periodic-molecules=yes" in my mdp file.
However, when attempting to use grompp, I get many errors of the type "No
default Bond types" "No default U-B types" etc.

Below is a snippet of the first residue of my structure:
    0ALA      N    1  -2.130  11.857   5.885
    0ALA     HN    2  -2.194  11.930   5.907
    0ALA     CA    3  -2.193  11.725   5.916

And a snippet of the last residue:
  179THR   HG23 2660   3.379  11.851   6.387
  179THR      C 2661   3.583  11.711   6.162
  179THR      O 2662   3.514  11.783   6.087

I have identified and added ~30 bonded interactions to my topology file.
For example, the carbon in THR should be bonded to the nitrogen in ALA, so
I have added the following line (line #5541 in my .top file):
[ bonds ]
;  ai    aj funct            c0            c1            c2            c3
2661 1 1

However, grompp returns the following error:
ERROR 1 [file monomer_3.top, line 5541]:
  No default Bond types

Similar errors are produced for most (but not all) of the added
interactions. All of the "pairs" and four of the "dihedrals" were processed
without errors. I cannot find a pattern in the dysfunctional vs functional
lines. For example, the following dihedral did not return an error:
(179THR      N 2649   )-(179THR     CA 2651   )-(179THR      C 2661 )-(0ALA
     N    1  )
However, this dihedral did:
(179THR     CA 2651   )-(179THR      C 2661   )-(0ALA      N    1  )-(0ALA
  HN    2  )

Any help in debugging this would be much appreciated!

Thanks in advance,

Rachel Baarda