Thank you for enlightening me and sorry for my ignorance,
Peter.

On Sun, Apr 25, 2010 at 5:33 PM, Frank von Delft <
[email protected]> wrote:

>  Hi, I sense a misunderstanding:
>
> Cell parameters are not relevant when scaling together datasets:  what
> matters are intensities and (depending on algorithm) directional cosines and
> the like.
>
> Cell parameters are relevant, along with the other experimental parameters,
> for knowing where to pick the intensities off the images ("integrating").
> What denzo & scalepack do (which I suspect you're referring to) is to split
> this into separate actions:  denzo integrates, and scalepack postrefines,
> meaning it optimizes the integrated intensities by simultaneously taking
> into account experimental parameters over whole dataset.  Only then does it
> scale.
>
> Xscale and Scala trust the intensities and only scale.
>
> Yes, the cell parameters are also an indicator that you may have
> non-isomorphism, but only that;  it's the *intensities* where this is
> manifested.  So you could have "different" cells yet still perfectly
> isomorphous intensities;  or "same" cells and terrible non-isomorphism.
>
> phx.
>
>
>
>
>
> On 25/04/2010 16:05, Peter Grey wrote:
>
> Dear Fred and Phil,
>
> However there is no refinement of these parameters in XSCALE so if you need
> to scale together several crystals (e.g. very small crystals that show
> severe radiation damage after a few degrees) you end up with a  sub-optimal
> combined dataset after XSCALE. I thought SCALA can take care of that and
> define cell parameters that fit best all data as scalepack does when it
> considers all separate frames , from all crystals where-as the input to
> XSCALE is not separate frames but complete datasets).
>
> Peter.
>
>
> On Sun, Apr 25, 2010 at 12:14 PM, Frederic VELLIEUX <
> [email protected]> wrote:
>
>> Hi Phil,
>>
>> Indeed it does, both during integration (INTEGRATE, only if you "ask for
>> it" with the proper keyword) and in the post-refinement step (CORRECT).
>> Normally this should be quite sufficient. I haven't seen a single case where
>> this was not sufficient.
>>
>> Fred.
>>
>> > Message du 25/04/10 09:55
>> > De : "Phil Evans"
>> > A : [email protected]
>> > Copie à :
>> > Objet : Re: [ccp4bb] SUMMARY: scala and xds data
>>
>> >
>> >
>> > Just to point out that Scala does not refine cell parameters: I believe
>> XDS does
>> >
>> > Phil
>>
>>
>
>
> --
> Peter
>
>


-- 
Peter

Reply via email to