Hi Vincent,
my advice is: if in doubt, scale at the highest of those resolution choices. This way you don't have to go back to scaling later. There is no harm in using the highest resolution. Don't let yourself be fooled by Rmerge.
best,
Kay
On Thu, 19 Jul 2012 17:52:14 +0200, vincent Chaptal <[log in to unmask]> wrote:
>Dear all,
>
>I'm dealing with anisotropic diffraction of a membrane protein.
>I've read from previous threads that it is better to not cut off any
>data, and that aimless doesn't anisotropically scale the data any way.
>So my question is: given my anisotropic diffraction, what resolution
>cutoff do i give aimless for scaling, and what statistics can i report?
>(log bellow)
>
>- Do I scale at 2.85A knowing that there is at least data in one
>direction to this resolution? (The different Rs are all over the place,
>but i can site the Karplus Science paper for the use of CC for
>resolution cutoff)
>- Do I scale at 3.47A as it is "the best overall estimate" (with decent
>R statistics).
>
>##### Aimless log #####
>
>Estimates of resolution limits: overall
> from half-dataset correlation CC(1/2) > 0.50: limit = 3.47A
> from Mn(I/sd) > 2.00: limit = 3.83A
>
>Estimates of resolution limits in reciprocal lattice directions:
> Along 0.98 h + 0.18 l
> from half-dataset correlation CC(1/2) > 0.50: limit = 3.97A
> from Mn(I/sd) > 2.00: limit = 3.65A
> Along k axis
> from half-dataset correlation CC(1/2) > 0.50: limit = 5.12A
> from Mn(I/sd) > 2.00: limit = 5.23A
> Along -0.25 h + 0.97 l
> from half-dataset correlation CC(1/2) > 0.50: limit = 2.85A
> from Mn(I/sd) > 2.00: limit = 3.17A
>
>################
>
>
>Thank you for your input.
>vincent
>
>
>
>Le 4/25/12 5:40 PM, George Sheldrick a �crit :
>> I think that anything that irrevocably modifies the experimental data
>> should be avoided whenever possible. Since anisotropic scaling is a
>> relatively fast calculation and there are several ways of doing it, it
>> is better to apply it locally when it is needed, e.g. in phasing
>> (where it is applied by phaser and shelxe etc.) and refinement (with
>> refmac or phenix_refine etc.). Provided that the standard deviations
>> of the observed intensities are properly taken into account,
>> anisotropic data truncation is not so important (i.e. as usual I agree
>> with Garib and Phil).
>>
>> George
>>
>> On 04/25/2012 06:19 PM, Phil Evans wrote:
>>> You can get the aimless documentation from
>>>
>>> ftp://ftp.mrc-lmb.cam.ac.uk/pub/pre/aimless.html
>>>
>>> pending its official release through CCP4
>>>
>>> No it does not do anisotropic scaling as such. That needs some sort
>>> of model of the "ideal" intensity, probably best calculated from a model
>>>
>>> I'm not sure that anisotropic cutoffs are a good idea. I believe
>>> Garib thinks they are not and I generally defer to him
>>>
>>> Phil
>>>
>>> On 25 Apr 2012, at 17:00, Bryan Lepore wrote:
>>>
>>>> wondering if aimless performs anisotropic scaling or "elliptical"
>>>> rejections lately.
>>>>
>>>> I ask because:
>>>>
>>>> [1] last I knew, scala did not
>>>> [2] I can't seem to google up the aimless manual as readily as scala
>>>>
>>>> ... also, what consesquence would mosflm anisotropic resolution limits
>>>> have on scaling (if aimless anisoscaling were true).
>>>>
>>>> -Bryan
>>
>>
>
>--
>
>Vincent Chaptal, PhD
>
>Institut de Biologie et Chimie des Prot�ines
>
>Drug-resistance modulation and mechanism Laboratory
>
>7 passage du Vercors
>
>69007 LYON
>
>FRANCE
>
>+33 4 37 65 29 01
>
>http://www.ibcp.fr
>
>
>
|