Wed Jul 21 13:19:45 PDT 2004
T0245

DUE 24 Aug


Thu Jul 22 14:12:50 PDT 2004 Kevin Karplus

No strong hits in t2k.best-scores, and scattered weak hits.
Similar results for t04.

try1 looks terrible, with the strands curled up into helices.


Fri Aug  6 11:25:06 PDT 2004 Kevin Karplus

On FORCASP (Date: August-3rd/04), Alexey Murzin wrote

    The category of this target is changed, as the coordinates of a
    homologous protein structure, 1TLJ, have been released today. The
    conserved hypothetical protein SSo0622 from Sulfolobus solfataricus
    has a 30% sequence identity to the target, making it the first
    suitable parent for homology modeling.


Fri Aug  6 13:40:14 PDT 2004 Kevin Karplus

I added 1tljA to the template library, and am doing
	make -k REDO_SEARCHES=1
to redo the searches.  With any luck, this will pick up the 1tljA
sequence and allow better modeling, so that we don't need to use
new-fold methods.


Fri Aug  6 14:33:19 PDT 2004 Kevin Karplus

1tljA is definitely coming up as the top hit now.
try2 will be exactly the same as try1 (so a fully automatic
prediction), just run after 1tljA was added to the library.


Fri Aug  6 21:33:56 PDT 2004 Kevin Karplus

The automatic try2 model looks pretty good.  It is very similar to the
model from the alignment to 1tljA, but has swung the N-terminal helix
a bit to bury the dry side of the helix better and to cover some of
the beta sheet.  This moves the conserved W3 close to conserved
residues R49, C46, H71, and I98.  I suspect that C46 and H71
coordinate a metal ion.

I think we may just need to do a polishing run and declare this done.


v
Sat Aug  7 12:29:30 PDT 2004 Kevin Karplus

I just noticed that the t04 fragments are not being created, and fixed
the Make.main file to ensure that they are.  It is not clear that they
would help any with this target, since we have a close homolog.

Try3 is looking pretty good.  It scores well with both the try3
costfcn and the unconstrained costfcn.  Rosetta prefers the
try2-opt2.repack-nonPC model, because it sees worse clashes in try3.

I could do one more polishing run with the unconstrained costfcn, to
try to fix up any remaining breaks or clashes.


Sat Aug  7 16:20:47 PDT 2004 Kevin Karplus

try4 makes a small improvement over try3 in the try4 costfcn, which is
an unconstrained cost function.  The result is just a bit loose.  I
should turn up the dry weights and turn down soft clashes a bit to try
to get tighter packing.

Rosetta still prefers try2, which it thinks has fewer clashes.


Sat Aug  7 20:07:09 PDT 2004 Kevin Karplus

We've reached the point of diminishing returns for this target.
I think it is time to submit
	try5-opt2
	try5-opt2.repack-nonPC
	try2-opt2
	try2-opt2.repack-nonPC
	T0245-1tljA-t2k-local-str2+CB_burial_14_7-0.4+0.4-adpstyle5