Ding ding ding ding...we have a winner:
- A search model with 30% identity was submitted to the Phyre
server which returned a target sequence-corrected model
- The Phyre monomer was "dimerized" in Pymol using symmetry
operators from a previous MR solution using a dimer of the
original homolog protein
- The Phyre dimer was used in Phaser to find a new MR solution
using the dimer (4 dimers placed, as expected, packs well in
unit cell.)
- This solution was Chainsawed to poly-Ala and then subjected to
refinement with 8-fold NCS symmetry averaging in Refmac to
generate phases. R=0.466, Rfree=0.460, FOM=0.219
- This model was fed to Parrot for 5 rounds of density
modification with 8-fold NCS symmetry averaging, improving FOM
to 0.770
- Parrot phases were fed into Buccaneer for 5 rounds of
autobuilding using the full target gene sequence and strict
sequence probability matching. The final Buccaneer model had
R=0.313, Rfree=0.345, and FOM=0.848, and the maps are readily
interpretable.
Omitting the Chainsaw step resulted in a much less satisfactory
solution, although perhaps usable, with R=0.41 and FOM=0.75, but the
map was not as interpretable, and had many problem areas and
possible mis-registrations.
My second choice of approach was going to be DM with phase extension
(Keller et al., 2006, Acta Cryst., D62, 1564-1570) which was
suggested by David Lawson. The MR challenge described in that paper
is eerily similar to mine. I may yet try this for kicks to see how
well it works ( may need this trick someday) but was too lazy to
feed all the NCS matrices into DM. Parrot will extract NCS operators
directly from the MR model, which is nice. :)
Thanks for all the very helpful advice!
On 9/2/2010 7:10 AM, Dima Klenchin wrote:
1. Chainsaw the currrent solution, and
attempt to identify and build in the correct register of the
side chains. after refinement.
2. Do a low resolution refinement of the poly-Ala/Gly model to
better thread the main chain?
3. Try EPMR with the Phyre model or the poly-Ala/Gly model
4. Give up and get real phases (but I'm so close now!?)
Roger,
Here is what I'd do if I were you:
1. Try both with and without Chainsaw.
2. Refine current solution with tight NCS averaging.
3. Feed the resulting phases into Parrot.
4. Feed the output to Buccaneer.
Since Phaser built "correct" tetramer, I'd be very surprised if
the above procedure won't resolve registration issue.
Good luck,
Dima
--
Roger S. Rowlett
Professor
Department of Chemistry
Colgate University
13 Oak Drive
Hamilton, NY 13346
tel: (315)-228-7245
ofc: (315)-228-7395
fax: (315)-228-7935
email: [email protected]
|
