Hi Mousheng,

[if it gets too SHARP-centric we can also move this discussion also to
sharp-discuss or our developers mailing list sharp-develop]

On Mon, Oct 18, 2010 at 08:33:47AM -0500, Wu, Mousheng wrote:
> my solvent content is supposed to be 65% (one molecule per
> ASU).

which seems to match your lowish resolution limit (as a general rule).

> Sharp automatically did density modification using 37.5% solvent
> content.

I see. Did you get a clear and significant difference in score for the
two hands? It not only tells you which is the correct hand, but a
large difference also tells you that one set of phases is
significantly better than the other and that there is real phasing
information in those.

> that is why I run solvent flattening directly from Sharp.

Did you use the tools within our interface? Probably ...

> >More confusion: what do you mean with "anomalous map"? One might call a map with anomalous differences and some phases such an "anomalous map" ... but I guess you rather mean an electron density map with phases from some procedure based on anomalous differences, right?
> 
> you are right. the anomalous map I calculated is anomalous differences map using FFT 

Ok - so you are comparing an anomalous difference map with an electron
density map? I'm not sure why, but:

 * the anomalous difference map (using phases from SHARP or your
   density modified phases) should only show the substructure,
   i.e. only the Se atoms and no light atoms

 * the electron density map after density modification shows
   everything: light and heavy atoms.

Why are you comparing those two maps? To check if your heavy atom
sites are where they should be or if all are correct?

If you trust your density modified phases (and you should, since they
show nice helical density): just calculate such an anomalous
difference map using your density modified phases ... it's a two
mouse-click operation in the SHARP/autoSHARP interface and should show
you peaks and their distance to your initial heavy-atom sites.

Are those 6 'poor' sites in any way different after SHARP refinement?
Much higher B-factor or lower occupancy? Do you have additional peaks
in your log-likelihood residual maps (showing the true position of
those potentially wrong sites or if they are split)? Do you have peaks
in the ANO residual maps exactly at the sites - showing wrong f"
values (similar for f' values in ISO residual maps)? I.e.: is your HA
model as good as it could be?

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