Dear Tim,
As I see it, the issue is not how good the overall correlation is, but whether the differences in distribution of the electrons (narrow vs. broader) will show up in difference maps, which often show up at the edge of the ion. To make such calculations, I would calculate a difference map between say a model with a Zn with full occupancy and high B-factor vs. a model with a Zn with say 80% occupancy and a matching lower B-factor.
Only if you do not see any significant features is such a map, I will be convinced there is not a great difference in changing occupancy or B-factor. Besides, it is new to me that increasing the B-factor would reduce the number of electrons. I always assumed they only would get distributed differently.
Best,
Herman
-----Ursprüngliche Nachricht-----
Von: CCP4 bulletin board [mailto:[log in to unmask]] Im Auftrag von Tim Gruene
Gesendet: Mittwoch, 7. Mai 2014 16:52
An: [log in to unmask]
Betreff: Re: [ccp4bb] AW: [ccp4bb] Refining Metal Ion Occupancy
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Dear Bernhard,
I just happen to collect the correlation between ADP and occupancy for a publication I am involved in.
At 1.5A (!) resolution, the correlation for a single ion between both figures is greater than 90% - there is certainly not a clear difference between these factors.
One of the reasons might actually be visualised from the URL you
posted: At 2.5A resolution (the resolution this thread is about) the number of electrons for Zn with B=30 drops from 30 to 25, which is not so great a difference, at 1.5A it drops to about 20, which is still not so great a difference, i.e. the B-factor weight is not too far off from being constant at 'protein' resolution ranges.
Best,
Tim
On 05/07/2014 02:58 PM, Bernhard Rupp wrote:
>> the negative difference density surrounding your metal ion shows that
>> the lower occupancy could not be fudged by a higher B-factor
>
> Because there is a clear difference between high B-factor and low
> occupancy: High B factor attenuates high resolution scattering most,
> while lower occupancy just evenly scales the scattering curve down.
> Ergo, the FT - the Electron density - also looks different, with a low
> occupancy causing a WIDER scattering curve than a comparable high
> B-factor, thus transforming into a NARROWER peak compared to high
> B-factor.
>
> So, you could adjust (within physically meaningful limits) B and n to
> 'reshape' the electron density. If you have a negative difference
> density 'ring', your 'observed' density there is less than the model
> density, and by reducing n you could reduce the wings of the model
> electron density peak, thus achieving a better match.
>
> There is also the possibility that you have - perhaps in addition -
> some truncation ripples, which are most prominent around heavy atoms.
>
> Figures 9-6 and 9-5 BMC. This app allows to generate the different
> scattering curve shapes, and a similar app lets you FT it.
> http://www.ruppweb.org/new_comp/scattering_factors.htm
>
>
> Best, BR
>
> -----Original Message----- From: CCP4 bulletin board
> [mailto:[log in to unmask]] On Behalf Of
> [log in to unmask] Sent: Mittwoch, 7. Mai 2014 14:25 To:
> [log in to unmask] Subject: [ccp4bb] AW: [ccp4bb] Refining Metal
> Ion Occupancy
>
> Dear Chris,
>
> In my experience, modern refinement program manage quite well to
> deconvolute occupancy and B-factor. In your case the negative
> difference density surrounding your metal ion shows that the lower
> occupancy could not be fudged by a higher B-factor. I would just
> refine occupancy and B-factor at the same time and let the refinement
> program do the deconvolution. If your density maps would still
> indicate problems, you always can try to manually deconvolute.
>
> By the way, your formulation <attempt to "flatten" the negative
> density> sounds like some cheap trick, when in fact you try to get
> a model that more accurately reflects your observed diffraction
> pattern.
>
> Best, Herman
>
>
> -----Urspr ngliche Nachricht----- Von: CCP4 bulletin board
> [mailto:[log in to unmask]] Im Auftrag von Chris Fage Gesendet:
> Dienstag, 6. Mai 2014 19:03 An: [log in to unmask] Betreff:
> [ccp4bb] Refining Metal Ion Occupancy
>
> Hi Everyone,
>
> In my 2.5-angstrom structure, there is negative Fo-Fc density
> surrounding a metal ion after refining in Phenix. From anomalous
> diffraction I am certain of the metal's identity and position in each
> monomer. Also, the ion is appropriately coordinated by nearby side
> chains. Should I be refining the occupancy of the ion in attempt to
> "flatten" the negative density? I am considering soaking the metal ion
> into crystals or cocrystallizing and collecting additional datasets.
>
> Thanks for your help!
>
> Regards, Chris
>
- --
- --
Dr Tim Gruene
Institut fuer anorganische Chemie
Tammannstr. 4
D-37077 Goettingen
GPG Key ID = A46BEE1A
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)
Comment: Using GnuPG with Icedove - http://www.enigmail.net/
iD8DBQFTakicUxlJ7aRr7hoRAq9bAKCZURDNZxvfzuXpk0slF1vh6hkkLACfd5Ls
s7OV453Zj63k1xMOo+qzTfA=
=lpSS
-----END PGP SIGNATURE-----
|