***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***


Hi, Vineet.

(If you've got an R factor of 45% already, why not try some rigid
body refinement on H1-110, H111-220, L1-110, L111-220 and have a look at
the maps? )

Flip Hoedemaeker's suggestion that you split a FAB into dimers is a
good one. Here's what I do:

1) From the PDB, select a FAB structure of the same isotype that I'm
working with. This gives 99-100% identity on the constant portion and ~60%
identity on the variable portion.
2) Take residues 1-110 from chains H and L, save the file as variable.pdb.
2a) Sometimes I cut out the CDRs, especially CDR-H3, since they're
unlikely to align well.
3) Take residues 111-220 from chains H and L, save the file as const.pdb.
4) Using MOLREP's ccp4i interface, find the best solution(s) for
const.pdb.
5) Fix const_molrep1.pdb and find the best solution(s) for variable.pdb.
6) View the result PDB file in PYMOL or COOT to ensure that the 110-111
transitions are in the right place (modulo symmetry operators).
7) Rigid body refinement, then look at unweighted maps.

Good luck!

-Anna Gardberg

-- 
Anna S. Gardberg, Ph.D.
Postdoctoral Research Associate
University of Tennessee


On Tue, 5 Sep 2006, Flip Hoedemaeker wrote:

>Hi Vineet,
>
>1) This might work better with Phaser.
>2) The interface between the two variable domains and the two constant
>domains is more rigid, so you could use two dimers, might improve the MR
>signal.
>3) In my memory (but I have not used it) there is an elbow.inp routine in
>CNS, check out the documentation.
>4) What is your resolution? Below 2.3 A you could throw your best solution
>in ARP/wARP...
>
>Hope this helps,
>
>Flip
>
>-----Original Message-----
>From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of
>Vineet Gaur
>Sent: Tuesday, 05 September, 2006 8:27
>To: [email protected]
>Subject: [ccp4bb]: molecular replacement for Fab molecule
>
>
>Hi,
>I'm trying to find a molecular replacement solution for a Fab molecule. Have
>tried a few models with AMoRe. None of them seem to have worked as yet. The
>best soln. till now has a correlation of 40 and a R factor of 45. I was told
>that this problem sometimes occurs because of variable elbow angles in the
>Fab molecules and consequent variable heavy and light chain domain
>interfaces. So I split my model in to 4 fragments - Light and heavy chains
>and each chain into variable and constant domains. Now used four models as
>input (to try n simulate a hetro-tetramer) in AMoRe but the program gets
>stuck at a stage with the following error message:
>'fixed part coming from a  6 body traing output
>orientations coming from a 6 body traing output
>stop >>o2tndo << no solution selected
>stop >> amore<< error in previous step'
>would somebody please help me figure this out. Any suggestions to solve the
>Fab molecular replacement problem in general??
>Thanks
>
>
>
>

Reply via email to