Although the CASP organizers haven't told us that this target is a dimer, this doesn't mean that it is a monomer.
As you noticed, the server models for R0008 were quite similar to one another, that is because R0008 is the only CASP ROLL target so far that has some remote homologs, which the servers we able to use as templates.
It turns out that some of these remote templates form dimers, which explains why the Zhang/QUARK server models had a crazy helix sticking out into nothing.
Obviously when you fold that up as a monomer, the helix can't stick out (that makes no sense) but you can see in the attached image that it makes sense in the context of a dimer:
In red is QUARK Model 1 superimposed onto the solved structure of a known protein in green & cyan (that was obviously used as a template by the QUARK server). As you can see, that template forms a dimer and we believe that this CASP ROLL target might do the same.
Since most of the 10 QUARK/Zhang server models were similar to one another, we selected the 3 different topologies across those 10 models so that you can focus on forming dimers with 3 models (instead of all 10).
___Some possible hints___
Because these are non-Rosetta server models, you might want to pull them apart first to clean them up before working on optimizing the symmetry.
Since this target has 128 residues, symmetry shouldn't slow it down too much. I still recommend initially turning off all sidechains, using Cartoon Thin with Hydro Color and only showing Clashes.
If you notice that the protein is scoring better when both subunits are far apart from one another, you'll need to find a more favorable interface between them.
Unfortunately this CASP ROLL target is due on March 8th, so it needs to close late on the 7th.
We hope you enjoy it even more than the Bonus R0007 Symmetry Puzzle!