My question may be interesting not only for me, but also for everyone who try to fold protein ab initio.
The question is:
Is it reasonably to make manual selection of predicted models in terms of reasonable topology and packing of secondary structure elements?
Let me explain. Oftren, as a result of ab initio we see a set of structures. Some of them looks loopy (from loop region), but another seems reasonably arranged with hairpin, superhelix, helix bundle e.t.c. Is it true that if structure adopt reasonable packing, it is closer to real structure then structure looking less ordered (bearing in mind that our protetin is not intrinsically disordered). Is anybody know rules for manual selection of structures, like "right handed topology of helices packing is preferred over left handed"?