Output structures have wrong chirality for some residues

I ran Haddock for protein-ligand docking, and the output structures have the wrong chirality (D- instead of L- amino acid) for a handful of residues. The starting structures have correct chirality for all amino acids, and I have run these exact same files before on the Haddock webserver (but with different constraint files) and gotten docked results that did not have this issue. How is it possible that the chirality got inverted for some of the residues?

The chirality is checked and defined based on the input structures by calculating the improper angle for the CA atoms.

Are you sure you input structures have a correct CA geometry?

Can you share one of these so that we can test?

Thank you for looking into this! I am quite sure that it is not the input coordinate file, because I’ve run this same structure several times and not had this issue. I am unable to upload directly to the forum, so here is a link to the protein input file:

I reran the docking, this time keeping non-polar hydrogens in place, and it helped somewhat, but I still got two residues that had bad stereochemistry. I believe this may be due to my dihedral restraints I am uploading, as without the dihedral restraints, I have not had this problem. Here is a link to my dihedral restraints file:

If you don’t have the issue without the restraints then is it suspect and your restraints must be the origin of the problem.
I run the begin stage of HADDOCK on your model and it does not detect any D-amino acids, i.e. they are all defined as L.
Which means your restraints must be causing distortions.

You can easily check if this is the case by downloading a full archive from the run from the server, and control the chirality of the structures in the begin directory (i.e. before docking).
The header of the PDB files in the begin directory will also tell you which amino-acids have been detected to be in a D chirality.

Also check the models in structures/it0 since these are the result of rigid body docking. If the changes are induced by your restraints they should occur only at the it1/water stages.

Since dihedral angle restraints are intramolecular by definition, why not first refine your structure against those restraints before the docking.
Should be a more efficient strategy.

Also simply run some validation on your starting model to check its quality.
You seem to have quite some distortions (also in omega angles)

Thank you so much for your help in troubleshooting this problem, and all the suggestions!

Indeed, the starting structure has some imperfect bond angles because its coming from MD, and the structure was not minimized after production. Since I am refining the docked complex, I will simply dock without the torsion angle restraints, and then implement the dihedrals during refinement.

Thanks again for the prompt responses!