[sharp-discuss] Combining MAD with native

[Clemens Vonrhein]

Hi Derek,

the best way of combining it is to use scenario 2:

  C-1
    X-1
      W-1
        B-1
      W-2
        B-1
      W-3
        B-1
  C-2
    X-1
      W-1
        B-1

with:

  - Se-S in C-1

  - use the f'/f" values for your Se atoms in C-1 (i.e. from
    fluorescence run): there is no difference here no matter if you
    use Se or Se-S.

  - nothing in C-2

  - give the 3 wavelength in C-1 in the order they were collected
    (usually peak, infl, hrem)

You should also refine

  - global non-isomorphism (based on isomorphous differences) for all
    datasets

  - local non-isomorphism (based on isomorphous differences) for 2nd
    and 3rd wavelength of your MAD as well as native

  - global and local non-isomorphism (based on anomalous differences)
    for all your MAD dataset

I had very good experience with MAD+native. It all depends obviously
on the (non)-isomorphism, for which the cell changes might give an
indication (and the 17% R-factor you quote might be more interesting
broken down into resolution shells). So even if there is some
non-isomorphism, the native might give you some low-resolution phase
information - which helps density modification, which then gives a
better map. Every little helps ...

If you send me your current SIN file I can have a look at it (to
double check everything is fine).

Btw: a NISO_BGLO of 4.4 isn't too bad at all. And: what do you mean
with 'the native contributes nothing to the phasing'?

Cheers

Clemens


On Tue, Mar 20, 2007 at 11:28:09AM +0100, Derek Logan wrote:
> Dear all,
> 
> I would like some advice on how to combine MAD and native in SHARP.  
> I've never done this before. I've tried following the instructions in  
> the manual, with little success. Two scenarios are described there:
> 
> 1) Use SeMet in the derivative and S-Met in the native
> 2) Use (Se-S)Met in the derivative and nothing in the native.
> 
> In both cases I use the peak wavelength of the MAD data set as  
> reference. I have a set of 11 Se positions from a simple MAD  
> refinement. The orthorhombic cell dimensions are 75.47  128.94   
> 175.98 for SeMet and 74.95  127.25  175.21 for native. The MAD data  
> go to 2.8Å and native to 2.3Å. SCALEIT gives an isomorphous R-factor  
> of 17%, which is maybe a bit on the high side for Se...
> 
> In scenario 1, if I let the occupancies of the S atoms in the native  
> vary then they fluctuate wildly, as do the Se occupancies in the  
> derivative. If I fix the S occupancies to 1.0 then nothing much  
> happens to the Se occupancies, but the NISO_BGLO parameter for native  
> goes up to 4.4 and the native contributes nothing to the phasing. Am  
> I refining incorrectly or are the data just very non-isomorphous?  
> There is after all a 1.7Å difference in the b dimension...
> 
> In scenario 2 I have not gotten very far because I have no idea what  
> to put in as f' and f'' for a Se-S atom. Any suggestions?
> 
> I would like to know generally speaking what is the correct course of  
> action for combining native with MAD and what experiences people have  
> had in the past.
> 
> Thanks
> Derek
> --
> Derek Logan             tel: +46 46 222 1443
> Molecular Biophysics    fax: +46 46 222 4692
> Lund University
> Box 124, Lund, Sweden
> 

> _______________________________________________
> sharp-discuss mailing list
> sharp-discuss at globalphasing.com
> https://www.globalphasing.com/mailman/listinfo/sharp-discuss


-- 

***************************************************************
* Clemens Vonrhein, Ph.D.     vonrhein AT GlobalPhasing DOT com
*
*  Global Phasing Ltd.
*  Sheraton House, Castle Park 
*  Cambridge CB3 0AX, UK
*--------------------------------------------------------------
* BUSTER Development Group      (http://www.globalphasing.com)
***************************************************************