You probably have an atom naming issue, most likely hydrogens. Rosetta doesn’t know what to do with mis-labeled atoms so it makes its best guess, which for your proline residue is wrong. You should be able to exclude particular patches with the -chemical:exclude_patches flag. “-chemical:exclude_patches pro_hydroxylated_case1 pro_hydroxylated_case2” should do it. (Another option is to just edit the database to remove the hydroxyproline patches altogether.) That’s a bit inelegant, though, as Rosetta will still discard and rebuild the misnamed hydrogen atoms. The best solution is convert the input hydrogen atom naming convention to the one Rosetta uses, but just ignoring the patches should work perfectly well for almost all cases.
By any chance are you using an input structure from Charmm? You might want to take a look at http://www.rosettacommons.org/content/charmm-hydroxyproline-hydrogen-atom-id
.
