For T0200, our fold-recognition method came up with a beta sandwich,
but the alignment was poor.

When we added predicted contacts to our method, we kept forming a
TIM-barrel.  We have attempted to build both a sandwich model and a
barrel model, but prefer the sandwich model.

Model 1 is try19-opt2, the best sandwich we came up with.
We believe that the C-terminal hairpin should be part of the sandwich,
but did not have time to figure out exactly where it fits in.

We particularly liked the way that the conserved polars clustered on
the C-terminal edge of the sandwich, probably marking a metal-binding site.