[ccp4bb]: Summary of 'Twinning problems (again....).'

[Mark Roe <roe@icr.ac.uk>]

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

Dear All,

Thanks to the many people who replied to my question. The overwhelming 
majority were of the opinion that the data were not twinned and that Rano 
doesn't need to be greater than Rmerge for there to be a signal - so I stand 
corrected.  This was succinctly stated by Eleanor Dodson  - 

>Ranom is the differences between Mn(I+) and Mn(I-) and will decrease as
>you increase multiplicity and get better data.
>
>But Rmerge reflects the scatter about a mean and usually increases with
>multiplicity - that is why it is a pretty useless measure of data
>quality..
>
>So your original premise that Ranom< Rmerge means the data is twinned
>is the problem - not the data itself..

Tassos Perrakis suggests the use of XPREP to check the data. This was 
actually run on the first set of data that we collected while we were at the 
ESRF and it indicated that the data were around 40-50%. This is where the 
idea originally got into my head. Initially I discounted this result because 
everything else looked OK. But since I haven't been able to solve the 
structure with either SOLVE, SnB or SHELXD, I was beginning to think that 
maybe XPREP was correct. Can someone tell me where to get hold of XPREP? Is 
it only available through Bruker? 

Eleanor also points out the XPREP analysis will also be available in SCALA 
in the new year.

Patrick J. Loll pointed out that "A trivial (if unpleasant) possible 
explanation--the Se-Met residues are all disordered" something I had 
considered but rejected on the account that there are (meant to be) 10 Se 
atoms in the a.s.u.

Both Pete Dunten and Nick Keep pointed out that the Rmerge is quite high 
especially in the low resolution bins. This I had noted (and also the rather 
low I/sigI) which was part of the reason I think something funny is going on 
with the data.

Tassos also points out that truncate is for cases of merohedral twinning - 

>because they are for general cases of 'merohedral twinning'.
>You can have a variety of other nasty artefacts like hemihedral twinning
>and whatever. You could be able to see funny effects in truncate output 
>in the table listing h/k/l odd/even intensities. If odd intensities are 
>less or more than evens that is usually bad news. hemihedral twinning 
>can be seen by carefull examination of the diffraction images, as double 
>spots in higher resolution with preference along specific lattice 
>directions.

and further

>We had 3 years of that. A p21 disguised as c2221 which was hemihedrally 
>twinned p21 at the end (or so I like to think). What worked was getting 
>actually another protein... If the protein xtallises and shows some 
>non-standard (merohedral) twinning (which is usually due to a 
>high-symmetry shape of the molecule) I think it usually means that you 
>have two separate protein species that interconverting during 
>crystallisation and can both be incorporated to the lattice, since the 
>difference is small. In MutS, which is an ATPase, adding ADP together 
>with cutting 53-c-terminal residues did the trick.

This may be an important clue. The protein involved is mistargetted by 
mutants that make the protein temperature sensitive. These switch at around 
30C. So even at 15C there will probably still be some population of both 
forms - enough to screw everything up maybe.

Anyway to sum up - It's not merohedrally twinned. But (there's always one of 
those) I still can't solve it (I'm still convinced the data is a bit ropey 
for synch. data) and I'm still banging my head against the wall.

Moving on to trying ACORN,

Cheers and thanks for all the help,
Mark.

--
Dr. S. M. Roe,
X-Ray Laboratory Manager,
Section of Structural Biology,
Chester Beatty Laboratories,
The Institute of Cancer Research,
237 Fulham Road,
London.
SW3 6JB.

Tel. (+44) 020 7970 6047
Fax. (+44) 020 7970 6051

E-mail roe@icr.ac.uk

*****************************************************************
*  "You've achieved success in your field when you don't know   *
*  whether what you're doing is work or play".  Warren Beatty.  *
*****************************************************************