[phenixbb] Disappearing density?

Tue Aug 10 07:36:43 PDT 2010

Hi Kyle,

It's always worth questioning your data and your model, so yes I'd  
examine both carefully.

I am guessing that this is  molecular replacement solution. (It isn't  
as common to have a loop that disappears during model-building with  
experimental phasing because normally you would not build it in the  
first place unless the density was there. My suggestions would be  
similar in either case though.)

So (still guessing) your starting MR model had some parts that are  
correct, some that are not, and some that are close.  You carry our MR  
and place your model. You calculate a sigmaA-weighted map, probably  
after refinement. This map shows great density in many places, medium  
density in others, little density in some.  You rebuild the parts that  
are clear to you, but some are not so clear, including the helix and  
loop in question. You refine some more, and now you want to improve  
the density in this unclear region.

Then as you say below, you calculate an average kick map, and it  
doesn't look so good in this unclear region. You calculate an  
iterative build omit map and the density in this region is nearly gone.

A comment or two first:

1. Yes, it is possible that the presence of a symmetry element very  
near the problematic area can make the density noisier. Normally this  
happens only right at the symmetry element however, not even a few A  
away.

2. The averaged kick maps and iterative build omit maps are ways to  
remove model bias. In general (with some exceptions) they are not  
great ways to "improve" maps.  Generally your best maps for building  
are going to be your current 2Fo-Fc map (e.g., from phenix.refine) or  
a density modified map using model information (e.g.,  
overall_best_denmod_map_coeffs.mtz from autobuild).  The omit and kick  
maps are to help you determine whether you have overinterpreted your  
density.

So if this is all about right, here is my suggestion: try to see if  
there is any evidence whatsoever for density in this region.

   A. Go back to the MR solution (very beginning of this structure  
solution, before any refinement except rigid-body). As this model has  
not been refined, you can delete any portion, refine the rest, and  
there will be no model bias in the map.
   B. So: delete the region in question completely from this unrefined  
starting model.
   C. Then refine the model, calculate a map, and look at the map in  
the region of your helix and linker.  If there is density (even not  
great density) that clearly matches the structure of the helix and the  
linker, then you know for sure that this helix and linker belong  
there.  If not...perhaps they are disordered or somewhere else.
   D. If no density yet, you can , iterative build the model, do  
whatever you want with this model to make it better except do not put  
the helix and linker back in...and then look at the map in the region  
of the helix and linker. Still you have not biased the map, so if you  
can find them they are really there.  If you cannot see them...now it  
is time to call them "disordered" until you get further evidence.

I hope that helps!
All the best,
Tom T

On Aug 9, 2010, at 3:08 PM, Kyle Dolan wrote:

> Hi all,
>
> I attempted to improve the density around a helix and short linker  
> region using an average kick map but it didn't make things look much  
> better. I then tried creating a composite iterative-build omit map  
> from phenix.autobuild, but in this case the electron density around  
> the helix/linker in question largely disappeared! Is there another  
> approach I can use that might improve my maps, or should I start  
> questioning my data? The dataset has >95% completeness to 2.7 A so I  
> am very curious as to why I am missing the backbone now, where  
> before I was only missing the side chain densities. One thing I am  
> wondering is if the problem could have something to do with  
> symmetry; the part of the map I am looking at may be close to a two- 
> fold crystallographic symmetry axis and I have been told that these  
> regions tend to be very noisy and hard to model--comments?
>
> Kyle
>
> Kyle T. Dolan
> Department of Biochemistry and Molecular Biology
> The University of Chicago
> ktd at uchicago.edu
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb

Thomas C. Terwilliger
Mail Stop M888
Los Alamos National Laboratory
Los Alamos, NM 87545

Tel:  505-667-0072                 email: terwilliger at LANL.gov
Fax: 505-665-3024                 SOLVE web site: http://solve.lanl.gov
PHENIX web site: http:www.phenix-online.org
ISFI Integrated Center for Structure and Function Innovation web site: http://techcenter.mbi.ucla.edu
TB Structural Genomics Consortium web site: http://www.doe-mbi.ucla.edu/TB
CBSS Center for Bio-Security Science web site: http://www.lanl.gov/cbss

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20100810/50938f51/attachment-0003.htm>