Quoting "Phil Evans" <[log in to unmask]>:
>
> Can anyone explain what Zbyszek Otwinowski means by Chi squared?
If I understand properly, CHI**2 value as used in Scalepack is:
SUM(<I>-Ij)**2/SUMsigma(I)**2 (I have to use formula editor to write
it properly, but the idea is clear) and is useful in exhibiting and
detection of systematic errors.
Intuitively this value will be close to 1.0 if only counting
statistics contribute to errors in measurements of I.
Drawing "CHI**2" distribution as a function of various values such as
frame number, shell of resolution, intensity
of reflection etc. may show very interesting things related to the
status of systematic errors during data collection.
In the case of absence of systematic errors and radiation decay and in
presence of a stable X-ray beam these distributions will be
featureless and their value will be dependent on a counting
statistics, quality
of a detector and absorption of a crystal mainly.
Systematic errors of various sources change these distributions in a
sensible way. Despite the fact that it is impossible to correct
systematic errors, understanding of CHI**2 distributions lead to
better understanding of the experiment and in most cases these
systematic errors can be eliminated leading to a "perfect" data shaped
mostly by a counting statistics.
"CHI**2" distributions a la Otwinowsky - Minor (or else, I also do
not know if Zbyszek Otwinovsky developed it by himself or adopted from
earlier sources),is as used in Scalepack an instrument of a great power.
FF
Dr Felix Frolow
Professor of Structural Biology and Biotechnology
Department of Molecular Microbiology
and Biotechnology
Tel Aviv University 69978, Israel
Acta Crystallographica D, co-editor
e-mail: [log in to unmask]
Tel: ++972-3640-8723
Fax: ++972-3640-9407
Cellular: 0547 459 608
> I can't find a definition in any of his papers (though I may have
> missed it). Is there a reference?
>
> It doesn't seem obviously related to the chi squared distribution
> ("In probability theory and statistics, the chi-square distribution
> (also chi-squared orχ²-distribution) with k degrees of freedom is
> the distribution of a sum of the squares of k independent standard
> normal random variables. "
> http://en.wikipedia.org/wiki/Chi-square_distribution)
>
> Phil
>
>
> On 29 Jun 2010, at 21:14, Felix Frolow wrote:
>
>> Graphical information from Scalepack as used in HKL2000 is of unprecedented
>> help to detect anomalous signal. Anomalous detection for S
>> anomalous data using
>> CHI**2 and Rfactor statistics for reflections with averaged and
>> separated Bijvoet pairs is attached. It is very well described in
>> HKL2000 manual. There is nothing special about data collection
>> (strategy was used) and measurement was relatively fast (4 h on
>> MicroMax007 and RaxisIV++).
>> BTW Rfactor is 1.9% with separated Bijvoet reflections and 2.6%
>> with averaged
>>
>>
>> BTW James Holton website calculate for this case 0.078 crystal
>>
>> Dr Felix Frolow
>> Professor of Structural Biology and Biotechnology
>> Department of Molecular Microbiology
>> and Biotechnology
>> Tel Aviv University 69978, Israel
>>
>> Acta Crystallographica D, co-editor
>>
>> e-mail: [log in to unmask]
>> Tel: ++972-3640-8723
>> Fax: ++972-3640-9407
>> Cellular: 0547 459 608
>>
>>
>>
>>
>>
>>
>>
>>
>> Quoting "Bernhard Rupp" <[log in to unmask]>:
>>
>>> I second the hkl2map/SHELXCDE approach. Two complete examples
>>> explaining how to do this for MAD and S-SAD cases are in my book.
>>> I wish to emphasize the importance of
>>> a) running enough trials
>>> b) careful selection of resolution cutoffs
>>> c) look at the solution distribution
>>> d) play with SHELXE parameters.
>>> The hkl2map graphs are enormously helpful for this purpose.
>>>
>>> BR
>>> -----Original Message-----
>>> From: CCP4 bulletin board [mailto:[log in to unmask]] On Behalf Of Tim
>>> Gruene
>>> Sent: Tuesday, June 29, 2010 2:29 AM
>>> To: [log in to unmask]
>>> Subject: Re: [ccp4bb] measure of anamolous signal
>>>
>>> Dear Murugan,
>>>
>>> you can use the program hkl2map from Thomas Schneider, available at
>>> http://webapps.embl-hamburg.de/hkl2map/
>>> It's a graphical interface to the programs shelx c/d/e which are available
>>> from http://shelx.uni-ac.gwdg.de/SHELX/index.html
>>>
>>> With SAD data you want to look at the d"/sigma line at the end of
>>> the shelxc
>>> output. Where that drops below about 1.3 is approximately where your
>>> anomalous signal ends. You might get slightly improved statistics
>>> with xprep
>>> instead of shelxc, but xprep is not free and you have to get a copy from
>>> Bruker-AXS.
>>>
>>> Tim
>>>
>>> On Tue, Jun 29, 2010 at 02:35:29PM +0530, Vandu Murugan wrote:
>>>> Dear all,
>>>> I have collected a 2.7 angstrom home source data with Cu-Kalpha
>>>> source for a protein with 6 cysteines, with a multiplicity of around
>>>> 23. I need to know, is there any significant anamolous signal present
>>>> in the data set, since there is no good model for my protein. Can any
>>>> one tell, which program to run, and what parameter to see? Thanks in
>>> advance.
>>>>
>>>> cheers,
>>>> Murugan
>>>
>>> --
>>> --
>>> Tim Gruene
>>> Institut fuer anorganische Chemie
>>> Tammannstr. 4
>>> D-37077 Goettingen
>>>
>>> GPG Key ID = A46BEE1A
>>>
>>
>>
>>
>> Dr Felix Frolow
>> Professor of Structural Biology and Biotechnology
>> Department of Molecular Microbiology
>> and Biotechnology
>> Tel Aviv University 69978, Israel
>>
>> Acta Crystallographica D, co-editor
>>
>> e-mail: [log in to unmask]
>> Tel: ++972-(0)-3640-8723
>> Cel: ++972-0547-459-608
>> Fax: ++972-(0)-3640-9407<detectionOfAnomalousSignalBy
>> Scalepack-HKL2000.pdf>
>
>
Dr Felix Frolow
Professor of Structural Biology and Biotechnology
Department of Molecular Microbiology
and Biotechnology
Tel Aviv University 69978, Israel
Acta Crystallographica D, co-editor
e-mail: [log in to unmask]
Tel: ++972-(0)-3640-8723
Cel: ++972-0547-459-608
Fax: ++972-(0)-3640-9407
|