Does comparative modeling works for membrane proteins

Dear all,

Does comparative modeling works for membrane proteins?
I found a application called "topology broker" in the documentation on "TopologyBroker_GPCR" included in "protocol_capture-3.3" bundle. According to the abstract, this protocol has been used for modeling of metabotropic glutamate receptor subtype 5 protein using GPRC crystal structures. But the "flags" is only for rebuilding loops of a membrane protein whose crystal structure is known.

Does anyone know exactly how to use this?

Thanks!
-Justin Zhang