Thu May 25 09:48:15 PDT 2006
T0302
Make started Thu May 25 09:48:51 PDT 2006
Running on orcas.cse.ucsc.edu

Thu May 25 16:25:24 PDT 2006 Kevin Karplus

pdb.blast picks up 2crpA with fantastically good scores (55%id over
129 residues, E-value 4e-39).  Blast finds about 13 templates.

The HMMs favor 1fqiA,but 2crpA is in undertaker-align as model 7.
try1 favors 1fqiA.

The top models from undertaker-align are all in agreement with try1.
We have very few clashes or breaks, so we might want to increase the
dry terms to try to pack things a bit tighter.



Tue May 30 09:15:20 PDT 2006 Kevin Karplus

I accidentally stepped on the Makefile and README file this morning
with new-target.  I restored them from my home computer, but I have to
fix new-target to avoid this problem.

Mon Jun  5 15:11:55 PDT 2006 Kevin Karplus

In a pattern that is becoming very familiar, the top-scoring model is
SAM_T06_server_TS1, then try1-opt2, variants on these two models, then
ROBETTA_TS3, ROBETTA_TS4, ...

All these models are very close, and very close to the alignments
(except the alignment to 2a72A).

I think I should do a polishing run from the SAM_T06_server_TS1 and
try1 models, then submit.


Mon Jun  5 19:19:41 PDT 2006 Kevin Karplus

try2-opt2 does score better than try1-opt2, reducing both clashes and breaks.
Rosetta actually like it after repacking.

I'll submit

ReadConformPDB T0302.try2-opt2.pdb
ReadConformPDB T0302.try1-opt2.pdb

ReadConformPDB T0302.undertaker-align.pdb model 1	(1fqiA)
ReadConformPDB T0302.undertaker-align.pdb model 2	(2bt2A)
ReadConformPDB T0302.undertaker-align.pdb model 3	(2a72A)

The 2a72A alignment seems to be a domain-swapped dimer.
We'll include it just in case that is what the target does.



Tue Jul 11 11:55:39 PDT 2006 Kevin Karplus

REAL_PDB is 2h33A

In my evaluations, our server did about as well as the ROBETTA server,
but ranked 30th among servers.  I'm porbably putting too much weight
on missing atoms, though as quite poor models that are complete are
beating better models that are missing a few atoms.

Our model 1 is slightly better than the server model, but nothing to
get excited about.

Fri Jul 14 11:51:11 PDT 2006 Kevin Karplus

With the improved evaluate.unconstrained.pretty

best server:	Zhang_Server_TS1	-1.26
our best model:	try2-opt2.repack-nonPC	-0.93
best submitted:	model1			-0.92
our best server: SAM-T99_AL4-scwrl	-0.91
SAM_T06_server:	31st of 54		-0.88