Thu Aug 15 14:56:16 PDT 2002
t0189

homology model 1rkd (c.72.1.1)

1 sept 2002 Kevin Karplus 13:00

current best model is try1-opt-scwrl.

There are a few bad breaks, and the C-terminal strand seems to be
detached from the sheet.  We may need to play with the alignment to
close the mid-strand break at V226-V227.  Perhaps we chould pair E230
with D234?

Let's do another run without constraints first, to see if any of hte
gaps close by themselves.


2 sept 2002 Kevin Karplus 19:49

The current best is try2-opt., which still has bad gaps.

Let's add anti-parallel strand F112-V317 next to V242-I245

237>   regVWLIRskeeiDSHLlg
319<  kVREVTFhdye


Let's add constraints to align A58 parallel to T82

52>  lgvpSVATGFvg
78>    sklitTNFVYVEge

4 Spet 2002 Kevin Karplus 21:19

I noticed that Oscar had started a try4 run, but that it had failed
because Template.atoms was not group-writable.  I made it
group-writable and started the try4 run over again for him.
It seems to be using the try3 constraints still, so I'm not sure it
will do much good. (CloseGap is turned up a bit high---that may be ok,
since we may be looking for just the sort of minor moves near the
break that Close Gap provides, but I suspect we really want to make
moves a little further from the gaps than CloseGap provides.)
We'll see how try4 does and look again in the morning.

We need to look at the structure and (probably) adjust the constraints
to make a more consistent request.  We could also lower the constraint
weight and increase the break weight to get better closing.

I put together a "conserve" rasmol script to show the conserved
residues.  The polar conserved residues all cluster nicely around a
pocket---presumably the binding site for the ligand.


Thu Sep  5 17:17:21 PDT 2002	Kevin Karplus

I looked at try4.0.100 (the current best score).

It is only slightly better than try3-opt (if that).

Let's do a try5 run, with a little more hope of getting some gap closing.

8 Sept 2002 09:58 Kevin Karplus

try5-opt made a little progress, not in closing the gaps but in
settling residues to pack a bit more tightly.  Let's up constraint and
break weights a bit, and do another run.  If that doesn't help much,
we may as well submit it as is.

15:42  try6-opt has larger breaks, but improved constraints.
Let's score everything WITHOUT constraints also.

20:41	try7-opt best without constraints, with try7 constraints, 
and with try3 constraints.

I'm not sure how much point there is to doing more---I'll submit try7,
with try6 and try5 as alternatives.