Pdb2gmx creating unphysical number of disulfide bonds per sulfur atom

GROMACS version: 2018.4
GROMACS modification: No

pdb2gmx does not check if the involved residues were already connected by a ss bond during the automatic usage of specbond.dat. (multiple peptide chains in a pdb file separated with “TER” should be crosslinked via disulfide bridges).
This leads to a topology file which contains angles like SG-SG-SG and dihedrals of CB-SG-SG-SG (SG is the sulfur atom in the CYS residue). There are obviously no parameters for these type of angles, so grompp throws error messages of missing interaction types.
Is there any solution to this? Did I miss somewhere an option making sure pdb2gmx only ever connects two sulfur atoms with each other?

pdb2gmx doesn’t know anything about chemistry. If it finds to S atoms in Cys within the distance tolerance set in specbond.dat, it links them. You will have to manually assign disulfides with -ss in this case since you have an odd geometry.

Thanks, so I guessed. But whats the point of automatic then if there is no check?

This is the first report I’ve ever seen of such an issue in 15 years of using GROMACS :)

There are always corner cases that are unaccounted for. But in a typical experimental structure, you’ll almost never run into this, and the option is there for the user to manually correct the issue in the extremely rare event that it arises.