While we're at it, it would also be a great improvement if the PDB would list nominal resolution in all 3 directions for structures with significantly anisotropic data. I'm never 100% sure what to do in those cases.
Thanks,
Phoebe
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Phoebe A. Rice
Dept. of Biochem & Mol. Biol. and
Committee on Microbiology
https://voices.uchicago.edu/phoebericelab/
On 3/1/20, 5:19 PM, "CCP4 bulletin board on behalf of dusan turk" <[log in to unmask] on behalf of [log in to unmask]> wrote:
Hi again,
First, I wish to thank everyone for their responses. I hope that no one minds that I include them in my response letter to the Editor ?
The idea of citing the Karplus and Diedrichs paper in Science has been essentially consumed already in our first response letter, which accompanied the submission of the revised manuscript, and did not work. (Instead of the Science citation from 2012 we used a quote from the paper suggested below by Karplus and Diederichs, Curr Opin Struct Biol. 2015 Oct; 34: 60–68. )
As Phoebe mentioned, it would be good to use the momentum. My intention of writing to the bulletin board was not only to get help with argumentation, but also to raise the issue of how to address resolution cutoffs in crystallographic data in publications to avoid such situations in general. I also think that the issue is more complicated than the last shell criterion with I/Isig > xx, Rmerge < yy, or cc1/2 > zz, or ...
1. Number of reflections in a shell effects the numbers significantly. The larger numbers of the shells there are the larger the numbers will be. The number of reflections in a shell depends also on the highest resolution and unit cell size. Hence we have a dependance of potential criteria on several parameters.
2. Refinement and data processing programs use different numbers of shells and even different ways of calculating shells. REFMAC typically uses 20 equal volume sliced shells), PHENIX is more complicated, as far as I understand (shell number may depend on the number of TEST set reflections in individual shell, shells can be defined according to equal slicing volume, some kind of log dependency or even linearly according to real space resolution), in MAIN there are 20 shells by default, but one can choose any of the mentioned slicing rules.
I suggest that we use this discussion to shape up guidelines that can later proposed for consideration to the IUCr committee for macromolecules. I prefer soft as opposed to strict borders. In the end, the structures do not speak for themselves, but are a mean to support one or more biologically relevant conclusions.
???
best wishes,
dusan turk
> On 29 Feb 2020, at 01:00, CCP4BB automatic digest system <[log in to unmask]> wrote:
>
>
> Date: Fri, 28 Feb 2020 16:03:22 +0000
> From: "Phoebe A. Rice" <[log in to unmask]>
> Subject: Re: What resolution - X-ray diffraction round this time
>
> Can we get some momentum for the "standard table 1" including TWO numbers - outer limit used in refinement, and nominal resolution based on some standard such as I/sigI =2 (or 3, or whatever the community can agree on)? That would hopefully cut down on all the reviewer complaints of overstated resolution.
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Phoebe A. Rice
> Dept. of Biochem & Mol. Biol. and
> Committee on Microbiology
> https://voices.uchicago.edu/phoebericelab/
>
>
> On 2/28/20, 6:56 AM, "CCP4 bulletin board on behalf of Malý Martin" <[log in to unmask] on behalf of [log in to unmask]> wrote:
>
> Dear colleagues,
>
> I agree with all the previous responses, it is a pity to throw away
> useful high-resolution data. The problem of high-resolution cutoff
> estimation is also nicely summarized in another paper by Andrew Karplus
> and Kay Diederichs "Assessing and maximizing data quality in
> macromolecular crystallography"
> https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4684713/ . It is suggested
> using CC1/2 for the selection of the cutoff for data processing (not
> I/sigI or R_whatever). Later on, the decision should be validated
> performing the paired refinement protocol.
>
> Good luck with the argumentation.
> Martin
>
>
> On 2/28/20 11:08 AM, LMB wrote:
>> Ask the referee - (apart from the other suggestions here)
>>
>> ‘How would removing data Improve my model?”
>>
>> Sent from my iPad
>>
>>> On 28 Feb 2020, at 08:22, dusan turk <[log in to unmask]> wrote:
>>>
>>> Hi,
>>>
>>> Browsing through the recent discussion on EM data resolution cutoff
>>> it occurred to me that the X-ray diffraction community isn’t that
>>> unanimous either.
>>>
>>> My stand:
>>>
>>> When the default resolution cutoff provided with the data processing
>>> software in electron density map calculation and refinement delivers
>>> quality maps noisier than expected and/or too high R-factors I start
>>> adjusting the resolution cutoff by lowering the resolution and trying
>>> alternative space group. Hence, I allow the data processing programs
>>> to suggest where to draw the line (be it CC1/2, I/sigI, R merge, R
>>> sym, R p.i.m. and R r.i.m, …) , unless there are problems.
>>>
>>> Doing so, I came into a dispute with a referee who shaped his request:
>>>
>>> "It is well accepted that the criteria for resolution cutoff should
>>> consider both I/SigI and Rmerge for the outer most shell. For data
>>> sets collected at synchrotron sources, the criteria of I/SigI > 5 and
>>> Rmerge <50% can be taken as a good practical reference.”
>>>
>>> So where do we stand? Which are the most objective criteria for
>>> resolution cutoff to be used in diffraction data processing? Which
>>> number of shells to use when calculating the statistics? Do we have a
>>> consensus?
>>>
>>> best wishes,
>>>
>>> dusan turk
>>>
Dr. Dusan Turk, Prof.
Head of Structural Biology Group http://stef.ijs.si/
Head of Centre for Protein and Structure Production
Centre of excellence for Integrated Approaches in Chemistry and Biology of Proteins, Scientific Director
http://www.cipkebip.org/
e-mail: [log in to unmask]
phone: +386 1 477 3857 Dept. of Biochem.& Mol.& Struct. Biology
fax: +386 1 477 3984 Jozef Stefan Institute
Jamova 39, 1 000 Ljubljana,Slovenia
Skype: dusan.turk (voice over internet: www.skype.com
########################################################################
To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1
########################################################################
To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1
|
