[phenixbb] ENSEmble parameters when the model is an EM map

Randy Read rjr27 at cam.ac.uk
Fri Jul 24 01:51:38 PDT 2009


Dear Peter,

On the first point, it shouldn't be necessary to worry about the  
overall scale and temperature factor, because Phaser works with  
normalized data (from which the overall scale and B have been  
removed).  This effectively makes the average intensity as a function  
of resolution the same for the EM map and the X-ray data.  On the  
other hand, you have to give Phaser an effective RMS error (i.e. the  
RMS error that would give the same structure factor error for  
coordinates as you expect for the EM image) -- I think a value of 1/3  
to 1/2 of the resolution of the EM image would be a good enough  
guess.  This is needed to calibrate the likelihood targets, so that it  
is known how well the observed and calculated structure factors should  
agree for the correct solution.

I didn't realize that the EMDB doesn't check whether the magnification  
is given in the deposition.  That's an issue that should be addressed.

If you're talking about cryo-EM data, where you get a 3D density  
distribution, then I don't think that cut-off levels should be an  
issue.  If it's negative stain data, then that would be a different  
story.  However, I don't have a good feel for how well negative stain  
images can be used for MR.  If anyone has any experience, it would be  
good to hear about it!

As for the origin and the extent, these don't have to be particularly  
precise, so you could just approximate them by visual inspection of  
the EM map.  The extent is used to compute how large the rotation step  
can be (i.e. how much you can rotate before the edge of the object  
rotates a large amount relative to the resolution of the data), and  
that rotation is carried out around the centre that you specify.   
Also, the quality of the interpolation depends on getting these  
approximately right.  Roughly, the cell into which you put the object  
has to be large relative to the furthest distance of a point in the  
object from the specified centre, so if you specify the centre badly,  
you would need a larger cell to get accurate interpolation.

Best wishes,

Randy Read

On 23 Jul 2009, at 19:08, Peter Grey wrote:

> Dear Prof. Read,
>
> Thank you for the swift and kind reply. I would like to mention than  
> there are several  difficult issues in using EM maps. For example it  
> would be wise to use a temperature factor to sharpen/blur the map so  
> that the histogram of  amplitudes vs. resolution would look similar  
> to that of the measured structure factors (better for MR). In  
> addition one needs to determine visually the cut-off level for the  
> mask. Visual inspection is also required to determine the  
> magnification because many maps are deposited in the EMDB with  
> absolutely arbitrary magnification. EMDB is not run like PDB and  
> standartization is not really practiced (sadly) - All these  
> parameters should have been given by the microscopist.
>
> I am familiar with the Phaser example you mentioned but there the  
> problem of finding the origin and extent is easy  because you start  
> from a pdb file (to obtain the mask). The approach I suggested is  
> very primitive. Is there a program that can determine more precisely  
> the "box" that contains the particle and its centre of  mass ? It  
> should be easy becasue at this stage you already have the mask  
> (whose cut-off was determined by visual inspection) that dictates  
> what is "particle" and what is not in the map.
>
> Peter.
>
>
>
> On Thu, Jul 23, 2009 at 5:51 PM, Randy Read <rjr27 at cam.ac.uk> wrote:
> Dear Peter,
>
> There's a new Phenix GUI for Phaser that's under development, and  
> when that 's finished you'll be able to run all possible Phaser jobs  
> from the GUI.  In the meantime, it's probably easiest to use command  
> scripts.
>
> A number of the practical issues of using EM maps are the same as  
> using density cut out from an X-ray
> electron density map, e.g. making sure the cell is big enough (so  
> that the interpolation of the transform of the density works) and  
> knowing where the centre of the object is.  These are discussed on  
> our web page, at http://www-structmed.cimr.cam.ac.uk/phaser/density_as_model.html 
> .
>
> Assuming that your large P1 cell is at least 2.5 times as big in all  
> directions as the small cell, then your approach sounds reasonable.
>
> For EM maps, there's an extra issue.  It seems that there is  
> generally an uncertainty of several percent in the magnification  
> factor, so you will want to adjust that up and down -- steps of 0.5%  
> would probably be fine enough.  The easiest way to do this is to  
> scale the cell dimensions in the MTZ file containing the structure  
> factors derived from the EM map.  You could do that easily in  
> sftools, and then run searches with all the scaled MTZ files.
>
> Good luck!  I'd be interested in hearing how you get on, and don't  
> hesitate to ask any further questions.
>
> Best wishes,
>
> Randy Read
>
> On 23 Jul 2009, at 16:08, Peter Grey wrote:
>
>> Dear benevolent experts,
>>
>> I am trying to use Phaser for molecular replacement with EM-map as  
>> a model.
>> The EM map, originally in a small P1 cell, was put in large P1  
>> using CCP4's MAPROT and a mask that was defined by a certain cut- 
>> off level.
>> Could you please advise me how to derive the EXTEnt and CENTre  
>> parameters needed for ENSEmble building ?
>> Does the simple way of giving the size of the original P1 as the  
>> EXTEnt and its centre as CENTre sound reasonable ?
>>
>> I would be grateful for any advice and comments,
>>
>> Peter.
>>
>> _______________________________________________
>> phenixbb mailing list
>> phenixbb at phenix-online.org
>> http://www.phenix-online.org/mailman/listinfo/phenixbb
>
> ------
> Randy J. Read
> Department of Haematology, University of Cambridge
> Cambridge Institute for Medical Research      Tel: + 44 1223 336500
> Wellcome Trust/MRC Building                   Fax: + 44 1223 336827
> Hills Road                                    E-mail: rjr27 at cam.ac.uk
> Cambridge CB2 0XY, U.K.                       www- 
> structmed.cimr.cam.ac.uk
>
>
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://www.phenix-online.org/mailman/listinfo/phenixbb
>
>
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://www.phenix-online.org/mailman/listinfo/phenixbb

------
Randy J. Read
Department of Haematology, University of Cambridge
Cambridge Institute for Medical Research      Tel: + 44 1223 336500
Wellcome Trust/MRC Building                   Fax: + 44 1223 336827
Hills Road                                    E-mail: rjr27 at cam.ac.uk
Cambridge CB2 0XY, U.K.                       www- 
structmed.cimr.cam.ac.uk

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20090724/99bd6fda/attachment-0003.htm>


More information about the phenixbb mailing list