Mon Jul 22 11:30:54 PDT 2002
t0160

22 July 2002 Kevin Karplus

This is an obvious homology model with gapless alignment to 1grwA
(C.elegans major sperm protein), 3mspA (Ascaris suum motile major
sperm protein), 1mspA, 2mspA.  Even BLAST finds this homology easily.

The domain is monomeric and highly resistant to proteolysis, so the
predicted barrel seems reasonable.

I'm a bit worried that the secondary structure prediction doesn't seem
to line up with the secondary structure features in the try1-opt
model, but our alignments seem to be in agreement with everyone else's.

The packing looks unusually good for a homology model--let's see if it
gets any better on the try2 run, with more clash reduction.  Yes, the
scores are getting better.

23 July 2002 Kevin Karplus

try2-opt-scwrl is the new best scorer.  Packing still looks good, but
I'm still worried about 2ry structure prediction.

Mon Aug 19 16:59:25 PDT 2002 Kevin Karplus

With modernized score function, best score is for 
T0160.try1+1mspA-T0160-vit-adp...+T0160-1grwA-local-adpstyle5.pw.a2m.gz:1grwA.17.20.pdb,
but next best is try2-opt-scwrl.

I've added a "try-heavy" target to try to get an alignment that agrees
more with 2ry predictions, and will re-run from scratch.

Tue Aug 20 14:22:09 PDT 2002 Kevin Karplus

New best score is try3-opt.  It looks quite good.
Let's do one polishing run (with dry weights, clashes and breaks
turned up) and be done.  The modified scoring function puts
try3-opt-scwrl first, so let's start from both.


21 Aug 2002 Kevin Karplus

try4-opt is new best.  There is one small break (before A104) and one
tiny break (before G64), both in loop regions---the two other breaks
in try3-opt have been healed.  Packing is pretty tight, but I'll do
one more polishing run with higher dry5 and dry6.5 weights.

Wed Aug 21 10:57:44 PDT 2002 Kevin Karplus

try5.5.100 is new best (NOT try5-opt).
Will do one very short polishing run with OneRotamer set high.

Oops, that crashes with an assertion failure about conf->clashes()
existing on new conformation (presumably from OneRotamer).

18:10 Kevin Karplus

try6-try5-7.10.40 is new best, with almost all changes coming from OneRotamer
optimization steps.  The packing looks pretty good---I'll submit it
now, not bothering with further tweaking, though the final helix has
unwound a bit and lost some Hbonds.