My guess is that you’re running into a hydrogen atom naming convention issue. Rosetta uses the older hydrogen atom naming convention for PDBs. PDBs from other sources don’t necessarily.
The result of this is that Rosetta will ignore “misnamed” hydrogens, discarding their coordinates. It will then rebuild the “missing” hydrogens according to ideal geometry. This will happen regardless of Rosetta application, so it doesn’t really matter if you use the original pdb or the scored output pdb – Rosetta will transform them both to the same thing internally, anyway.
If the exact positions of the hydrogens matter to you, I’d recommend renaming them in your input PDB to match Rosetta conventions. (Most of the time the exact hydrogen positions don’t matter, and would probably need to be refined to fit into the Rosetta energy function, anyway.)
.
