[sharp-discuss] Extending 3.9A phases to 2.6

[Clemens Vonrhein]

Hi Frank,

On Fri, Nov 16, 2007 at 11:51:26AM +0000, Frank von Delft wrote:
> Hi
> 
> I'm sitting with a slightly tricky one here, Se-MAD phases to 3.9,
> native to 2.6 (peak,remote,native).  HA refinement runs very stably,
> substructure is complete and all that,

Are you doing 2-wvl MAD or MAD+native in SHARP?

> now I'm trying to get the phases extended to 2.6A, and wanted to ask
> if there tips for the best solvent flattening.

I've reordered your questions, since it makes more sense answering
them this way:

> What are the main parameters to play with?  I do a DM run first
> every time, because the sharp maps don't show much info either.

I usually don't run DM before going into SOLOMON: especially not if
you're still after that 'significant difference between hands' sign,
that shows you that you're on the right track (the DM runs mess up
statistics).

> 3) The envelope looks a bit cleaner for P4132 than P4332, but I
> don't really see protein-like features (the odd sheet or helix would
> be nice).

The envelope doesn't really tell you anything: after all, you told the
program(s) to flatten 50% (or whatever) of the cell ...

> 2) I don't see too much difference between the two hands, both for
> I(4) and I4/I1; does this mean I have not in fact solved the
> substructure, or just that the phase extension is struggling?

Not good! I hope you're looking at the SOLOMON statistics (no DM run!)
at Cycle 1 to compare the two hands? Have you checked that you're
using the right spacegroup when inverting the heavy atom sites?

How sure are you about the substructure? And about the number of
mol/asu (76% solvent for 3.9 seems sensible - but you have a native to
2.6, where I wouldn't expect such a high solvent content)?

> 1) Solvent content by matthews should be 76% (enough Se for 1 mol/ASU), 
> but I(4) looks better at 50%, as does I(4)/I(1) (which is what solvent 
> optimization uses).  Is this usual?

Anything in the self-rotation? Without a clear difference between the
enantiomorphs, looking at the stats to determine the best solvent
content is a bit futile I guess.

Have you got any homology model you could try and do phased molecular
replacement with: to check if it docks into one of the two hands
(density modified maps)? Or try BUCCANEER to build a few helices or
sheets?

If you haven't done so yet: try MAD+native in SHARP.

Cheers

Clemens

-- 

***************************************************************
* 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)
***************************************************************