If you’re using RosettaScripts, you can specify the fold tree with the AtomTree mover (https://www.rosettacommons.org/docs/latest/protein-interface-design-movers.html#AtomTree)
If you’re using another protocol, you’ll have to look for protocol-specific options to specify the fold tree. In general, there isn’t a way to specify an arbitrary fold tree. (Although some protocols will obey an input foldtree from a silent file, so you could create a silent file with a particular foldtree with RosettaScripts and the AtomTree mover, and then use that silent file as input instead of the PDB.)
.
