Thu Aug 15 14:56:24 PDT 2002
t0190


t0190	1etb1			homology modeling b.3.4.1

Thu Aug 15 21:24:29 PDT 2002	Kevin Karplus

try1-opt-scwrl looks pretty good.  Nothing blew up and there are just
a couple of breaks to fix.


1 Sept 2002 13:13 Kevin Karplus

try2-opt now best.  There is just one small gap to close and we're done.


16:25
try3-opt now best, but gaps still not closed---too much OptSubtree
and not enough CloseGap or InsertFragment.

19:22 best is now try4.0.150 (not try4-opt!).

The gap is probably about as closed as it is going to get, though an
optimizer that was willing to use phi-psi angels that weren't from
training-set fragments could probably close up the remaining gaps.

Let's submit.