We had strong hits for the central domain, but the N- and C-terminal
ends were done with essentially new-fold techniques.

We did not put much effort into this target, because we were swamped
with targets by the time we got around to working on this.  We had
initially classified it as "comparative", not noticing the size of the
unaligned parts, and did not allocate enough time for a new-fold
prediction. 

Thiis try3-opt2, another possiblity for the surrounding domain.