Does FlexOracle predict the hinge points in GPCR homology models? And has anyone done such a submission.
FlexOracle, as well as hNMb and stonehinge, assumes the protein is solvated -- this is less than ideal for you since GPCR is a membrane protein. FlexOracle works by splitting the protein into two fragments at every possible pair of points, and computing the stability of the fragments using the FoldX force field. However FoldX will not guess at the effect of lipids, it can only estimate the effect of a solvent environment. That having been said, I have had some luck with membrane proteins, if only because often people are interested in some cytosolic domain. This may be the case for you.
TLSMD, on the other hand, uses the crystallographic B-factors and so it will tell you how it flexes in the crystal, which is often similar to how it will flex in vivo. All of these analyses are done for you when you submit to the HingeMaster server on molmovdb.org, as I encourage you to do. Bear in mind that the analysis will be done on one chain only. For particularly large proteins you may have to wait a couple of days. If you don't hear back from me when you get your results you are welcome to email me and I will help you interpret the results.
Threaded models have an additional strike against them in that the structure is almost certainly not properly equilibrated and thus is likely to be full of residual strains. This further limits the ability of FlexOracle to probe stability of fragments. hNMb is less sensitive to such details and so may be more useful, especially for the cytosolic domains. TLSMD will actually not return any results, since there will be no B-factors in a theoretical structure.
It still costs nothing to make a submission to our server, and despite these various issues you may still learn something.