When Alexey Murzin announced on FORCASP the availability of 1t0t as a
template, we set aside all our previous work, added 1t0t to our
template library, and ran everything from scratch.  Not surprisingly,
1t0t turned out to be an excellent template.  

Our final model is quite similar to the template and to the
automatically generated model, with the main differences being at the
N- and C- termini and in the loop at Q117-K129.  

The C-teminal variations are trivial, but we had two slightly
different positions for the N-terminus.  In this model we have the
N-terminus capping the cleft that has a lot of conserved residue
without filling it.

We had some difficulty in closing the loop automatically, and
eventually ended up cutting and pasting between two models that had
flaws in different places in the loop, and reoptimizing the chimera.

Because of the very tight deadline, we did not attempt to optimize the
model as a multimer, so there are probably some poor rotamers, and
possibly even some backbone problems in the multimer interface.

This is try12-opt2.repack-nonPC, whose backbone was created by
undertaker (same as model 1), but whose sidechains were repacked by
Rosetta.  It had the lowest Rosetta energy of any of our repacked
structures.  Most of the differences in rotamer are for surface
sidechains.