is it possible to run ab initio folding for part of a protein while fix the rest part? Say, I want to fix two fragments to be beta strands and the rest are free to fold around the strands. Is loop modelling the way to go? The un-fixed part are not necessarily loops.
The homology_modeling_with_end_extension demo (in rosetta_demos) should be usable for what you want to do.
If you just want to fix that two sections _are_ beta sheet, but not a particular conformation, you can do that via their fragments (sheet-only), and constraints to force those two sheets to interact.