I am sorry for the alleged mission creep, but I wrote this mail before the 
other responses to the thread except the one re the misleading target goal, 
but it went out with quite some delay thanks to COX broadband....
- there was no reference intended in 'recipes'  to the later posts, and indeed, I intended 
to make the initial response as broad as possible given the little information available.

(inside joke: blame it on Table 1)

Best, BR

-----Original Message-----
From: CCP4 bulletin board [mailto:[log in to unmask]] On Behalf Of Gerard Bricogne
Sent: Tuesday, March 15, 2016 12:04 PM
To: [log in to unmask]
Subject: Re: [ccp4bb] on final R value

Dear Bernhard,

     I think you may rather have "drowned the fish" (my best attempt at translating a French expression) by broadening the context too much.

     Smith had a very specific question, namely: how can it happen that such a low resolution structure has such a good R-factor? No matter what R-values are worth or not in general, I was just pointing out that a specific procedure (targeting to a known high-resolution
structure) has been developed to fulfill a very specific role:
preventing a model of a protein-ligand complex from "unrefining", compared to a known ligand-free structure for that protein, during a refinement performed against lower-resolution data for that complex, e.g. because the DMSO used to dissolve the ligand has lowered the diffraction limit of the crystals. This approach is very useful, e.g.
to get better difference density for the ligand, but will clearly be the source of systematically abnormal combinations of R-value and resolution - for perfectly understandable reasons.

     Mentioning this procedure as a plausible cause for what Smith was puzzling over seemed a reasonable idea, as he had quoted an intention of preventing "over-refinement" - which I thought was another name for what I was calling "unrefining", the idea being to stop refinement from doing damage.

     Finally, pointing out that the statistics of structures refined with targeting will be outliers from the common trends is another reasonable remark. They are *intended* to be so, precisely because they have been prevented from unrefining against only their own data, but it would seem important to have some kind of clear markup for them so that they do not confuse subsequent data mining.

     I wasn't trying to say anything of broader significance :-) .
     
     
     With best wishes,
     
          Gerard.

--
On Tue, Mar 15, 2016 at 10:34:50AM -0700, Bernhard Rupp (Hofkristallrat a.D.) wrote:
> The optimization of overall restraint weights has been discussed 
> multiple times on ccp4bb, and while
> 
> the details depend on your specific case, two things are not in dispute:
> 
> (a)   the absolute value of R (free) at a given resolution is *not* a hard
> target value, but depends on your molecule, crystal, protocol, 
> skills...The distribution is broad, cf. figures in various papers and BMC.
> 
> (b)   rmsd of bonds ~0.02 A is the upper limit, derived from small molecule
> crystal structure data bases, and *not* a target value.
> 
> 
> 
> The more you rely on geometric restraints (low resolution, less data), 
> the tighter that overall rmsd distribution generally becomes. The 
> resulting rmsd at lowest Rfree (or -LLfree) at the point of refinement 
> convergence should be meaningful (that is, not to exceed ~0.02A) and 
> the rest again depends on resolution, your molecule, crystal, weight 
> optimization protocol, skills…
> 
> 
> 
> https://www.jiscmail.ac.uk/cgi-bin/webadmin
> 
> 
> 
> There is no single rigid recipe/protocol for refinement, because every 
> structure is different. Most clear and simple recipes how to tackle 
> complex problems are generally wrong…..
> 
> 
> 
> Best, BR
> 
> From: CCP4 bulletin board [mailto:[log in to unmask]] On Behalf Of 
> Smith Liu
> Sent: Tuesday, March 15, 2016 6:37 AM
> To: [log in to unmask]
> Subject: Re: [ccp4bb] on final R value
> 
> 
> 
> Thanks Eugene.
> 
> 
> 
> The paper I read was from a top journal. The resolution was 3.9, and 
> the R-factor was 0.233. I was interested to know how to get Rfactor 
> 0.233 from
> 3.9 A resolution map.
> 
> 
> 
> Smith
> 
> 
> 
> 
> 
> 在 2016-03-15 19:42:48，"Eugene Osipov" <[log in to unmask]> 写道：
> 
> 
> 
> Dear Smith,
> 
> R-factors not the aim but indicators. My teacher taught me that with 
> correct weighting scheme rmsd of bonds and angles in your model should 
> be similar to rmsd's of monomers in library. So I try to keep weight 
> which gives rmsd of bonds ~0.02 after refinement. Focus rather on 
> correct description of your model in terms of chemical and physical sense.
> 
> 
> 
> 2016-03-15 10:57 GMT+03:00 Smith Liu <[log in to unmask]>:
> 
> Dear All,
> 
> 
> 
> I just read a sentence " To prevent over-refinement, an appropriate 
> weighting of the geometry versus the crystallographic term was 
> established empirically, aiming at good model geometry with a low R 
> value". By CCP4 refmac refine, will you please let me know which 
> strategy are helpful for getting lower R value?
> 
> 
> 
> Smith
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> --
> 
> Eugene Osipov
> Junior Research Scientist
> Laboratory of Enzyme Engineering
> A.N. Bach Institute of Biochemistry
> Russian Academy of Sciences
> Leninsky pr. 33, 119071 Moscow, Russia
> e-mail: [log in to unmask]