Unfortunately, the protocols exposed on ROSIE are intended to be simplified “canned” protocols for particular use cases. They don’t necessarily have all the adjustments which permit fine tuning of protocols. As far as I’m aware, the protocol currently exposed on ROSIE is not able to model nanobody structures.
I do believe that the local installs of Rosetta can be adjusted to model nanobodies, but it takes a bit of fiddling and is not something we’ve (yet) exposed to the ROSIE interface.