I am trying to model circular permutation in my protein which is a beta
barrel and I modeled it the following way--
AAAA-----BBB
--------CCC-------
If this is really your alignment, Modeller will try to use structural
information from both B and C for two residues (and nothing at all for
the linker between A and B). That's probably not going to work. You
might want to manually perform the mutation first in one of your
template PDB files, and give Modeller that modified template.