[phenixbb] Phenix.refine on coordinate-optimised structures

[Pavel Afonine]

Hi Joey,

your question can accommodate multiple interpretations and thus multiple 
answers as result. Dorothee's was one. Here is another one.

First off, what's the goal? Do you want to get a better model of your 
crystal structure or lowest possible R- factors?

Besides, depending on the data quality (resolution, completeness, etc) 
you'd need to choose a good parameterization of ADPs (individual 
anisotropic for all but H if resolution is sub-angstrom, isotropic plus 
TLS at medium to low, etc). Then also depending on data quality, you 
might be able to observe and model static (space) disorder using 
alternative locations. If that's the case you will want to refine 
occupancy factors of such partially occupied sites.

If the resolution is not low, are you sure you placed all convincing 
water? If not, you may want to let phenix.refine to do this for you 
automatically. And then towards the end make sure to check them in Coot.

There is a lot of choices and possibilities in-between, as one can imagine.

If you have more questions or need more help, we can possibly do it more 
efficiently by chatting over Zoom, let me know and I will set up a call.

Pavel

On 8/7/20 19:22, Dorothee Liebschner wrote:
> Hi,
>
> The strategy depends also on the resolution of the data, so if you 
> could tell us that, we can give suggestions. Also the type of data, I 
> guess X-ray diffraction?
>
> Out of curiosity, how do model quality metrics look like (clashscore, 
> ramachandran, etc) at this point? If you keep coordinates fixed, they 
> won't improve in refinement.
>
> Best wishes,
>
> Dorothee
>
> On Fri, Aug 7, 2020 at 2:52 PM Joey Farrell <jpfarrell at uchicago.edu 
> <mailto:jpfarrell at uchicago.edu>> wrote:
>
>     Hello,
>
>     I am trying to design a set of input parameters that I will be
>     passing through the phenix.refine function. The structures
>
>     that I am working with have already had their X,Y, and Z
>     orthogonal Å coordinates optimised via separate protocol.
>
>     I am looking for advice on which strategies and parameter values
>     to use with phenix.refine to improve the crystallographic R-factors
>
>     of such structures while preserving their earlier optimisation in
>     real space.
>
>     I have been having some success using “strategy = 
>     tls+individual_adp+occupancies” (where tls = “chain A”, tls =
>     “chain B”, etc.) as
>
>     well as adjusting the wxu_scale value until R-work and R-free are
>     minimised. Are there other strategies or phenix.refine
>     functionalities
>
>     that may also prove helpful with this end in mind? I believe there
>     may be ways to further improve R-work and R-free with this software
>
>     that I am not yet aware of.
>
>     Thank you for your help!
>
>     Best,
>
>     Joey
>
>     _______________________________________________
>     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>
>
>
>
> -- 
> Project Scientist, Molecular Biophysics and Integrated Bioimaging
> Lawrence Berkeley National Laboratory
> 1 Cyclotron Road, M/S 33R0345
> Berkeley, CA 94720
> Tel: (510) 486-5709
> Fax: (510) 486-5909
> Web: https://phenix-online.org <https://phenix-online.org/>
>
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb
> Unsubscribe: phenixbb-leave at phenix-online.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20200807/152bd129/attachment.htm>