Hi Rafael,
I guess that you ask for a number to be used as a "cutoff", but there is no such number. Completeness is measured on a continuous scale, typically from 0 to 100%. Higher is better, of course. Resolution shells with low completeness just carry proportionally less information about the model and its electron density. Using a cutoff, i.e. throwing away data of good quality e.g. in and beyond a resolution shell of "low" completeness, does not improve the resulting refinement nor the electron density - rather, it makes everything worse.
Ultimately it is the number of accurately measured reflections that counts, because each such reflection adds a certain amount of information (which depends on its intensity and sigma).
To get an idea about the gradual worsening of electron density with decreasing completeness, check out James Holton's movie at https://bl831.als.lbl.gov/~jamesh/movies/ . I believe that this synthetic example omits data equally in all resolution shells.
In practice, completeness is often a function of resolution. It helps the electron density to have high completeness at low resolution, where the strong reflections are.
Nowadays there is little reason for not having high completeness data sets, by simply collecting enough data.
My rules of thumb are:
a) For triclinic and monoclinic crystals, merging two or more sweeps with rotation of 180° around different axes, using a kappa goniostat, gives completeness >95%.
b) For higher (than tri- and monoclic) symmetry crystals, a single sweep typically gives data with >95% completeness.
c) Rotating more than 180° increases the multiplicity, which improves the precision of the data, unless radiation damage sets in.
Hope this helps,
Kay
On Thu, 29 Feb 2024 21:33:24 +0000, Rafael Marques <[log in to unmask]> wrote:
>Sorry for jumping into the post, but I would like the community’s opinion about completeness, once this topic was raised here. What could be considered reasonable? Recently I have seen a 65% completeness Crystal structure and, surprisingly, the electron density map was not that bad for a > 3.2 A structure. How such a nice map could have been calculated with such poor parameters? I could only think of anisotropy.
>
>Best
>
>
>Rafael Marques da Silva
>PhD Student – Structural Biology
>University of Leicester
>
>Mestre em Física Biomolecular
>Universidade de São Paulo
>
>Bacharel em Ciências Biológicas
>Universidade Federal de São Carlos
>
>phone: +55 16 99766-0021
>
> "A sorte acompanha uma mente bem treinada"
>________________________________________________
>
>________________________________
>De: CCP4 bulletin board <[log in to unmask]> em nome de Paul Adams <[log in to unmask]>
>Enviado: Wednesday, February 28, 2024 2:58:16 PM
>Para: [log in to unmask] <[log in to unmask]>
>Assunto: Re: [ccp4bb] Rwork and Rfree the same?
>
>
>By setting wxc (weight of the X-ray term) to 0.1 there is good chance that the refinement is dominated by the geometry term and the model isn’t really seeing the effect of the X-ray data. I suspect this would result in R-factors that are similar. Why they are so low is less clear, but as others have pointed out 38% completeness is a problem. It would be good to check if that is 38% overall, or just very incomplete in the higher resolution shells. If it is complete at lower resolution you might be able to do something with the dataset, but not using the default parameterization in refinement programs - you’ll need to apply constraints and additional restraints if you can, and look at the weighting (by modifying wxc_scale, not wxc).
>
>There is a Phenix mailing list you might want to use as well (I assume you are using phenix.refine): https://phenix-online.org/mailman/listinfo/phenixbb
>
>On Feb 28, 2024, at 8:21 AM, Justin Cruite <[log in to unmask]> wrote:
>
>Thanks everyone,
>
>I agree, 18.4% Rwork and Rfree is too good to be true for a 3.4 Å dataset. The data was processed using autoProc and the staranisano mtz was used for MR. The completeness is only 38%. It could be that the Rfree and Rwork reflection sets are small because of this? What is the best way to check the number of reflections used for Rwork and Rfree? Is this dataset usable at all?
>
>Thanks!
>
>Justin
>
>On Wed, Feb 28, 2024 at 10:21 AM nicfoos <[log in to unmask]<mailto:[log in to unmask]>> wrote:
>Hello Justin,
>
>There is something weird in your results. You mention Rwork/Rfree of
>0.1837.
>This means a pretty good refinement and also is very unusual to be
>obtain for a resolution of 3.37.
>Additionally you should not have Rfree = Rwork.
>I suspect something wrong with you Rfree reflections sets. What size is
>it ? Is your dataset complet ?
>How did you cut the res. ?
>
>I hope this may help you.
>
>Nicolas
>
>
>
>On 2024-02-28 16:10, Justin Cruite wrote:
>> Hi everyone,
>>
>> What does it mean if your Rwork and Rfree are exactly the same?
>>
>> I solved a 3.37 Å structure with Phaser-MR and immediately ran 10
>> cycles of refinement with wxc = 0.1. Everything else at default. The
>> Rwork and Rfree are both 0.1837. Is this bad?
>>
>> Thank you!
>>
>> Justin
>>
>> -------------------------
>>
>> To unsubscribe from the CCP4BB list, click the following link:
>> https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
>
>________________________________
>
>To unsubscribe from the CCP4BB list, click the following link:
>https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
>
>--
>Paul Adams (he/him/his)
>Associate Laboratory Director for Biosciences, LBL (https://biosciences.lbl.gov)
>Principal Investigator, Computational Crystallography Initiative, LBL (http://cci.lbl.gov)
>Vice President for Technology, the Joint BioEnergy Institute (http://www.jbei.org)
>Principal Investigator, ALS-ENABLE, Advanced Light Source (http://als-enable.lbl.gov)
>Laboratory Research Manager, ENIGMA Science Focus Area (http://enigma.lbl.gov)
>Adjunct Professor, Department of Bioengineering, UC Berkeley (http://bioeng.berkeley.edu)
>Member of the Graduate Group in Comparative Biochemistry, UC Berkeley (http://compbiochem.berkeley.edu)
>
>Building 91, Room 410
>Building 978, Room 4126
>Tel: 1-510-486-4225
>http://cci.lbl.gov/paul
>ORCID: 0000-0001-9333-8219
>
>Lawrence Berkeley Laboratory
>1 Cyclotron Road
>BLDG 91R0183
>Berkeley, CA 94720, USA.
>
>Executive Assistant: Michael Espinosa [ [log in to unmask] ][ 1-510-333-6788 ]
>Phenix Consortium: Ashley Dawn [ [log in to unmask] ][ 1-510-486-5455 ]
>--
>
>
>________________________________
>
>To unsubscribe from the CCP4BB list, click the following link:
>https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
>
>########################################################################
>
>To unsubscribe from the CCP4BB list, click the following link:
>https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
>
>This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/
########################################################################
To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/
|
