T0263 has moderately strong fold-recognition hits on ferredoxin-like
folds particularly the dimeric alpha+beta barrel.

We spent some time on the last day before the target was due, trying
to get the C-terminal piece to form an edge sheet on the other
monomer.  Because undertaker does not yet understand multimers, we had
to make a double-length sequence, optimize it, split it apart,
redimerize, merge the dimer into a double-length sequence, and iterate.

At one point we used DeepView to force a chunk of the final strand to
be straight so that undertaker could try to make the hbonds properly.
We did not have time to experiment with the register of the final
strand, and we don't believe that this first attempt at aligning the
strand is correct, and the awkward shape at L92-N96 could probably be
improved by getting the register right---we just ran out of time.  

We really need to improve undertaker so that it understands homodimers
correctly---optimizing the double-length chain provides no guarantee
that half the chain when re-dimerized will form the correct
dimerization interfaces.  Indeed, we ran out of time before we had
successfully created a monomer that properly closes the barrel when
dimerized.

This is from-dimer-try9B, the monomer from the second half of a
double-length chain.