[sharp-discuss] residual maps

[Clemens Vonrhein]

Hi Greg,

On Fri, Apr 24, 2009 at 07:01:28AM -0400, Gregory Bowman wrote:
> Hi,
>
> I have a question about residual maps. I have a low resolution (3.5-4A) 
> 2-wavelength SeMet dataset, and am refining the position/B/occupancy for 
> the 21 Se atoms. All of the Se atoms are basically correct and give 
> pretty good phases (FOM ~0.45)

I tend to ignore the FOM values and rather look at the phasing power
plot - especially at the trend in those:

 - do you get high values at low resolution and lower values at high
   resolution?

   * if you have values in the lowest resolution shell that are lower
     than the following ones: start worrying about your low-resolution
     data (beamstop, overloads etc)

   * if there is a big difference in those values given after the last
     refinement cycles (when outlier rejection is still on) and the
     final table (at the phasing, electron density stage - where all
     reflections are used): have a look at the outlier rejections
     (based on likelihood) to see if there is a pattern in rejected
     reflections, especially the ones with much lower log-likelihood
     values. If these are low-resolution reflections: overloads,
     beamstop masking etc.

 - at what resolution does the phasing power drop below 1.0?

   This might tell you what the map from SHARP will look like before
   density modification. If your data goes to 2A but this drop occurs
   at 4.5 you will have a hard time getting density modification to
   improve the phases all the way from 4.5 to 2A (especially if you
   don't have NCS to help).

> and reasonable backbone density. What is a bit confusing is that the
> residual map for the anomalous peak data gives positive density (>5
> sigma) on some atoms,

Your f" value for the peak is too low: check your fluorescence scan
again (I hope you did one _and_ kept it) or switch on refinement of
that parameter for the peak wavelength.

> whereas the anomalous high-remote data give negative density at some
> of the same atoms.

Same idea: f" for hrem is too big and wants to become smaller.

You can refine those two values simultanously, but leave all others
fixed (since you don't have an indication that other f' or f" values
might need refineing). Also: check that you haven't inadvertently
mixed peak and hrem by picking the wrong file ... sounds silly, but
happens easily.

> This occurs on five of the 21 atoms (for some of these the remote
> maps show positive density and the peak negative), and the other 16
> atoms are mostly OK. I'd appreciate any suggestions on how I might
> be able to further improve the HA model.

All of the above assumes that the refinement has converged and that
this density is more or less exactly at the HA positions (and not
slightly to the side - which might show anisotropy in the data or
sites ... or maybe split sites).

Cheers

Clemens

>
> Thanks,
> Greg
>
>
>
>
>

> _______________________________________________
> 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)
***************************************************************