[phenixbb] Appropriate model_vs_data usage over part of a structure

[Dale Tronrud]

   You can do real-space refinement against such maps, and we regularly
do inside of Coot, for example.  Such refinement should only be for big
changes to the model where reciprocal space refinement is of limited
utility.  The final word is to perform restrained, reciprocal space
refinement because, as you say, there is a clear distinction between the
information from outside sources (the restraints and the diffraction
pattern) and the model.  The situation is, mathematically, pretty
squishy when you are using as "observations" a map that is partially
derived from the model you are trying to improve.

   When you have an electron density map from microscopy the situation
is quite different.  Then the map is an independent source of
information and you don't have a dog chasing its own tail.

Dale Tronrud

On 5/25/2016 11:14 AM, Andy Watkins wrote:
> Thanks for clarifying its purpose! I guess I do still have one question,
> though: are ccp4 files computed via phenix.maps (from e.g., a .pdb and
> .mtz) unsuitable for real-space refinement (e.g., some kind of
> structural optimization with a scoring function that includes an
> electron density term) because such a ccp4 file is itself influenced by
> the starting model?
> 
> On Wed, May 25, 2016 at 1:57 PM, Dale Tronrud <detBB at daletronrud.com
> <mailto:detBB at daletronrud.com>> wrote:
> 
> 
>        An mtz file is a generic container for information that is
>     represented as Fourier coefficients - amplitudes and/or phases.  The mtz
>     you get from data reduction contains your observed diffraction pattern.
>     There are quite a variety of things that might be in there, e.g.
>     intensities, amplitudes, indications of anomalous scattering.
> 
>        The "map_coeffs" mtz probably echos all of those data and includes
>     the amplitudes/phases which can be used to calculate several maps.
>     These coefficients are derived from both your observations and the
>     coordinates of your model that matches the "map_coeffs" file.  The
>     program could simply write a map file directly, but the Fourier
>     representation takes less disk space and is quicker to read and write
>     than a map.
> 
>        The "map_coeffs" file is a "throw away" file.  You use it to look at
>     your map, but when you continue refinement you should always use the
>     original mtz.  You don't want to refine your model against an mtz that
>     contains data calculated from your model.  While it might be okay, it is
>     safer to stick with the original.
> 
>     Dale Tronrud
> 
>     On 5/25/2016 10:38 AM, Andy Watkins wrote:
>     > Thanks--that certainly makes sense. In that case, I suppose I have a
>     > different question: what distinguishes the .mtz file suffixed
>     > "map_coeffs," which is output by phenix.maps, from the input .mtz?
>     >
>     > On Wed, May 25, 2016 at 1:28 PM, Christian Roth
>     > <christianroth034 at gmail.com <mailto:christianroth034 at gmail.com>
>     <mailto:christianroth034 at gmail.com
>     <mailto:christianroth034 at gmail.com>>> wrote:
>     >
>     >     You might look for the Real space correlation coefficient to
>     see if
>     >     your model fit that particular map better than before, but thats
>     >     probably it.
>     >
>     >     There is no subset experimental data just influenced by one chain.
>     >     The whole complex contributes to each measurement and vice versa
>     >     your chain A model contribute to each measurement you made to
>     >     generate the maps. Sorry, but that's it
>     >
>     >     Christian
>     >
>     >     Am 25.05.2016 um 17:31 schrieb Andy Watkins:
>     >>     Perhaps necessary preface: I am largely learning as I go, and
>     >>     large parts of Phenix (and the details of refinement data file
>     >>     formats) are still black boxes for me.
>     >>
>     >>     Suppose I have a model of one chain of a complex and an .mtz
>     >>     describing the entire complex. I have performed some refinement
>     >>     procedure altering the model of that single chain.
>     >>
>     >>     My aim is to determine how this refinement procedure has affected
>     >>     the fit to the experimental data--or, really, the subset of the
>     >>     experimental data expected to be relevant for my model.
>     >>
>     >>     For obvious reasons,
>     >>
>     >>     phenix.model_vs_data chain_A.pdb whole_complex.mtz
>     >>     phenix.model_vs_data chain_A_refined.pdb whole_complex.mtz
>     >>
>     >>     are undesirable: they evaluate the model of a single chain
>     against
>     >>     the whole complex's data, and there's obviously a lot of
>     >>     unsatisfied electron density. So, while this command /works/, it
>     >>     reports an unrepresentative fit. (In theory, one solution to my
>     >>     problem would be to re-combine each chain's refined version, then
>     >>     run model_vs_data on the recombined, refined complex. I'm
>     >>     interested nonetheless in how it would work on the chains separately.
>     >>
>     >>     Now, during the refinement procedure, we of course generate .ccp4
>     >>     density maps for the individual chain models:
>     >>
>     >>     phenix.maps chain_A.pdb whole_complex.mtz
>     >>
>     >>     which produces the ccp4 file as well as, crucially,
>     >>     chain_A_map_coeffs.mtz. Attempting to employ that resulting .mtz
>     >>     file, i.e.
>     >>
>     >>     phenix.model_vs_data chain_A.pdb whole_complex.mtz
>     >>     phenix.model_vs_data chain_A_refined.pdb whole_complex.mtz
>     >>
>     >>     leads to an evocative error: Sorry_No_array_of_the_required_type:
>     >>     No reflection arrays available.
>     >>
>     >>     My highest suspicion is that I need to alter maps.params in a
>     >>     particular way so that reflection arrays are also output to
>     >>     model_map_coeffs.mtz. I could also imagine that I need to be using
>     >>     another program entirely!
>     >>
>     >>     Thanks in advance for whatever help you can provide;
>     >>     unfortunately, I can't provide any input files.
>     >>
>     >>
>     >>     _______________________________________________
>     >>     phenixbb mailing list
>     >>     phenixbb at phenix-online.org
>     <mailto:phenixbb at phenix-online.org>
>     <mailto:phenixbb at phenix-online.org <mailto:phenixbb at phenix-online.org>>
>     >>     http://phenix-online.org/mailman/listinfo/phenixbb
>     >>     Unsubscribe: phenixbb-leave at phenix-online.org
>     <mailto:phenixbb-leave at phenix-online.org>
>     >>     <mailto:phenixbb-leave at phenix-online.org
>     <mailto:phenixbb-leave at phenix-online.org>>
>     >
>     >
>     >     _______________________________________________
>     >     phenixbb mailing list
>     >     phenixbb at phenix-online.org <mailto:phenixbb at phenix-online.org>
>     <mailto:phenixbb at phenix-online.org <mailto:phenixbb at phenix-online.org>>
>     >     http://phenix-online.org/mailman/listinfo/phenixbb
>     >     Unsubscribe: phenixbb-leave at phenix-online.org
>     <mailto:phenixbb-leave at phenix-online.org>
>     >     <mailto:phenixbb-leave at phenix-online.org
>     <mailto:phenixbb-leave at phenix-online.org>>
>     >
>     >
>     >
>     >
>     > _______________________________________________
>     > phenixbb mailing list
>     > phenixbb at phenix-online.org <mailto:phenixbb at phenix-online.org>
>     > http://phenix-online.org/mailman/listinfo/phenixbb
>     > Unsubscribe: phenixbb-leave at phenix-online.org
>     <mailto:phenixbb-leave at phenix-online.org>
>     >
> 
> 
>     _______________________________________________
>     phenixbb mailing list
>     phenixbb at phenix-online.org <mailto:phenixbb at phenix-online.org>
>     http://phenix-online.org/mailman/listinfo/phenixbb
>     Unsubscribe: phenixbb-leave at phenix-online.org
>     <mailto:phenixbb-leave at phenix-online.org>
> 
> 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 196 bytes
Desc: OpenPGP digital signature
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20160525/c6fd7c9e/attachment.sig>