Mon Aug  5 19:19:11 PDT 2002
t0167

Clear homology to 1jeoA (c.80.1.3)

7 Aug 2002 Kevin Karplus

T0167.try1-opt-scwrl looks pretty good---even has pretty good
packing.  There are still a few breaks in mid helix, so maybe we
should reoptimize with the break penalty higher and lots of fragment insertion

Thu Aug 22 13:59:52 PDT 2002 Kevin Karplus

try2-opt is not bad, but has some breaks and could be denser.

I'm submitting it to the VAST server to get a structure-structure
alignment that may give us a cleaner alignment for positioning the breaks.
The search is VS20753 with password casp5t0167.
Hmm, the VAST server can't find that result.
Trying again,  VS30754 with password casp5t0167.
(Hmm, I must have misrecorded the first number--it should have been 
VS30753.)

I edited the VAST alignment slightly (at the termini and around the
two gaps), and am re-running the optimization with it as the only
initial alignment.  It will have some trouble filling in the loop
where there is an insertion, I suspect, but the results should be
better than try1 and almost as good as try2.  I'll then polish try3,
perhaps allowing crossover with try2.


22 Aug 2002 Kevin Karplus 21:13

try3-opt is new best score, easily beating try2-opt.
Let's polish it overnight and submit it in the morning.

Fri Aug 23 18:25:09 PDT 2002

try4-opt is new best.  It took longer to polish than I expected.
Let's submit it.