Unknown coordinates error in torsion angle dynamics step

I am running HADDOCK 2.2 on the milou server, docking two proteins (well, two heterodimers) using distance restraints from XLMS data (5 restraints at the moment).
One heterodimer is protein X plus a fragment of protein Y.
The other heterodimer is a second copy of protein X bound to a different fragment of protein Y.
I’m trying to dock these two heterodimers together.

I am defining several fully flexible regions.
I am finding that it0 proceeds and finishes OK, but then there is an error arising at the start of the torsion angle dynamics stage (missing coordinates listed for all atoms in one of the protein X copies):

%ATMCHK-ERR: unknown coordinates for atom "B -10 -ASP -N "

At the start of the dynamics, I also see a number of messages like this:

NBONDS: found 352064 intra-atom interactions
NBONDS: found 0 intra-atom interactions
NBONDS: found 7323 intra-atom interactions
%atoms "B -35 -VAL -CB " and "B -660 -MET -CE " only 0.42 A apart

On the side, I also notice a number of lines with this comment earlier on:

%PARRDR-info: duplication of bond C OC
[and similarly for many angles etc - seems like an issue with the parameter file?]

Sometimes the runs work though, so I don’t think there is anything wrong with my coordinates files.

I’m sure it’s something simple I’m not doing right…

Best regards

The error points to clashes inside your protein, which might lead to “explosion” of your structures.
The PARRDR error is not the cause of your problem.

“Bad” models do not give problems in the rigid body phase of HADDOCK since they are treated as rigid bodies. In it1 however the internal energies are also calculated and this can lead to problems.