I have a problem with scoring a pose after switching it back from centroid to fullatom representation. I did some troubleshooting and this is what happened:
1. Load PDB into PyRosetta
2. Score it with the 'standard' scorefunction (value ~ -43 REU)
3. Switch pose to centroid representation (do nothing else)
4. Switch pose back to fullatom representation.
5. Score pose again with same scorefunction in Step 2 (value ~35,0000 REU)
Why does switching back to fullatom representation destabilize the structure so drastically, even though I didn't perturb it in any way?