Hi Ben,
There's an example in the JCSG archive (2PBL, 47717) which exhibits
the same phenomenon, as a result of non crystallographic symmetry. For
example (from xtriage)
Acentric reflections
<I^2>/<I>^2 :3.669 (untwinned: 2.000; perfect twin 1.500)
<F>^2/<F^2> :0.504 (untwinned: 0.785; perfect twin 0.885)
<|E^2 - 1|> :1.144 (untwinned: 0.736; perfect twin 0.541)
Britton analyses
Extrapolation performed on 0.15 < alpha < 0.495
Estimated twin fraction: -0.247
Correlation: 0.9968
This is all explained a little further up:
The full list of Patterson peaks is:
x y z height p-value(height)
( 0.500, 0.500, 0.500 ) : 88.593 (4.638e-08)
Then in the digest further down...
The analyses of the Patterson function reveals a significant
off-origin peak that is 88.59 % of the origin peak, indicating pseudo
translational symmetry.
===
So. analysis is that the NCS gives rise to a near centring operation,
which makes the reflections with h + k + l even stronger and h + k + l
odd much weaker - this is the opposite effect to twinning, where the
reflections are "averaged" skewing the distribution to the middle of
the intensity range. In this situation the statistics which are used
to detect twinning don't work, which gives rise to the negative twin
fractions.
Short answer: read the whole xtriage output! You probably have
translational NCS reported in there somewhere.
Cheers,
Graeme
2009/9/22 Ben Flath <[log in to unmask]>:
> I have used phenix.xtriage. It finds three twin laws with alpha ~0.48 for all of them and the results of the perfect twin test are well over 2.
>
> Ben
>
> On 9/22/09 1:50 PM, Tanner, John J. wrote:
>> I suggest you analyze your data with phenix.xtriage. It does several tests and gives descriptive output.
>>
>>
>> ________________________________
>> From: Ben Flath <[log in to unmask]>
>> Reply-To: Ben Flath <[log in to unmask]>
>> Date: Tue, 22 Sep 2009 14:31:26 -0500
>> To: <[log in to unmask]>
>> Conversation: [ccp4bb] perfect twin test
>> Subject: [ccp4bb] perfect twin test
>>
>> Hi all
>>
>> when subjecting my data to the perfect twin test <I*2>/<I>*2 I get values very close to 3 which is far off the theoretical values of 1.5 and 2 for twinned and untwinned data.
>>
>> can anyone shed some light on what might be going on with my data. Could tetartohedral twinning have anything to do with it?
>>
>> Thanks
>>
>>
>> ---------------------------------
>>
>> Benjamin Flath
>> College of Pharmacy and Nutrition
>> University of Saskatchewan
>> 320 Thorvaldson Building
>> 110 Science Place
>> Saskatoon, SK
>> S7N 5C9
>
|