Lindsay,
At this time, the pmut_scan_parallel app can’t take a resfile (which would allow you to give residue-specific behaviors like NATRO). That’s an interesting idea, though. Perhaps I can add that in and have it as part of the next release.
Since you have an idea of which residues should be allowed to mutate and which to hold fixed, you can put that into a resfile, run a few dozen fixbb trajectories and then generate a sequence profile of the results. You can usually tell from that which positions Rosetta thinks are important for increasing stability.