[sharp-discuss] Model + HA on floating origin

[Clemens Vonrhein]

Hi Frank,

On Thu, Aug 09, 2007 at 06:37:50PM +0100, Frank von Delft wrote:
> Hi
> 
> We have the following case:
> 
> * Se-SAD dataset (3.3ish), substructure solves easily, but phases not 
> quite good enough.

good.
 
> * MR solution on same dataset from rather distant homologue (<25%ID), 
> phases also not quite good enough.

good.
 
> * MR solution aligns well (enough) with Se substructure, so both 
> solutions are correct.
> 
> We want to combine the two in Sharp.  However:  It's R3, so there is a 
> floating origin, and aligning a dodgy MR template with the Se sites is 
> tricky, or one uses the ccp4 jiffy.  But what about Sharp, is it bright 
> enough to spot this and optimize the various phase sets before it starts 
> refining?

Not SHARP - but you can use MOLREP. Get the latest from
http://www.ysbl.york.ac.uk/~alexei/molrep.html:

  http://www.ysbl.york.ac.uk/~alexei/downloads/molrep_linux.gz

and run it like this:

  1. create your best SHARP (or density modified) map:

       % fft hklin eden_flat_51.2pc.mtz \
             mapout eden_flat_51.2pc.map <<e
         LABI F1=FBshasol PHI=PHIBshasol
         e

  2. run MOLREP using

       % molrep -f eden_flat_51.2pc.map -m yummy_homologue.pdb

     At the end you should have a molrep.pdb file fitted/docked into
     the density. If you know already exactly how many molecules you
     have, you could also use

       % molrep -f eden_flat_51.2pc.map -m yummy_homologue.pdb -i <<e
         NMON 4

         e

     See 'molrep -h' for a good overview (and Alexei's web-page).

> (I ask because combining the non-optimized substructure and MR model 
> does not do wonders yet for the phases.)
> 
> 
> II. Does autosharp do anything magical such a substructure+model 
> combination, or does it really just cycle automatically what can be done 
> by hand?

autoSHARP doesn't have a channel for a user-supplied model. What it
does (in 'accurate' mode) at the end is cycling between density
modification and automatic model building (using ARP/wARP programs) -
including the latest available model from this step in density
modification.

In your case (after fitting the MR-model into density using MOLREP) I
would do the following:

  1. include the fitted model into density modification (for the first
     5-10 cycles only to avoid too much bias):

     - place the PDB file as model_something.pdb into
       sharpfiles/datafiles

     - declare yourself 'expert' (in the Preferences page on the SHARP
       Control Panel)

     - go to the 'Phase Improvement ...' panel of your latest/best
       (and in correct hand!) SHARP run (from the 'results' page off
       the SHARP Control panel)

     - start a density modification (optimizing) selecting the model
       (and tell it if it has the heavy atoms in it or not) and the
       number of cycles to use.

  2. you should get a better density modified map that was using SHARP
     phases + model phases for the phase combination during the first
     N cycles.

  3. If you already get a much better idea what is good/bad about your
     model: correct your model as well as possible and redo step 1

       - or -

     Use the tool on the 'Phase Improvement ...' panel to extract
     ABCDs from the latest/best density modification run for use in
     SHARP: this will create a new MTZ file (it tells you which name)
     that can then be used when running SHARP: select this MTZ file
     and clicking the boxes at the bottom of the General page (all
     excluding the final one). This way you will include the density
     modified SHARP+model phases as external phase restraint in the HA
     model refinement - giving you probably a better HA mode - and
     better experimental phases.

Obviously: always check your residual maps to see that everything is
looking fine and happy. Can you see additional peaks (for your Cys-S
atoms) in the ANO residual maps? This could help in tracing your
sequence as well ...

Hope that helps and gives you some ideas.

Cheers

Clemens

> 
> 
> phx.
> 
> _______________________________________________
> 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)
***************************************************************