Dear Almudena,
merohedral twinning does not give rise to "good" and "bad" spots. Rather, two lattices are exactly on top of each other, and contribute to each spot. From the data processing viewpoint, this is the same as if no twinning would be present. See http://strucbio.biologie.uni-konstanz.de/ccp4wiki/index.php/Twinning#Twinning_by_Merohedry
It sounds like you have two lattices _not_ always giving spots on top of each other, and some reflections overlap, whereas others don't. The lattices have a specific orientation relative to each other. This is called non-merohedral twinning. See http://strucbio.biologie.uni-konstanz.de/ccp4wiki/index.php/Twinning#Non-Merohedral_Twins
Finally there is the case of cracked or multiple crystals in the beam, where two or more slightly rotated parts give rise to two or more different lattices.
This would be not so easy to process. AFAIK there are specialized data processing programs for this; MOSFLM may be suitable - it can integrate several lattices at a time, whereas XDS can only integrate a single lattice.
HTH,
Kay
On Fri, 15 Jul 2016 12:22:47 +0200, Almudena Ponce Salvatierra <[log in to unmask]> wrote:
>Dear all,
>
>thank you very much for your suggestions. I will try. Last night I was
>doing some MR with Phaser and to some extent I succeeded, I did not do
>refinement yet though!
>
>@ Kay. Sorry!! I didn't mean to shoot anybody. :-) I just tried a quick
>thing with XDS to separate "bad" from "good" spots and then re-indexed the
>good ones, and eventually realized it wouldn't work with my data. But I was
>all in a rush and as I mentioned I never had to deal with twinning before,
>so.. accept my apologies if it sounded like it was XDS's fault! it was not!
>
>All the best,
>
>and thank you once again!
>
>Almu
>
>2016-07-15 9:31 GMT+02:00 Kay Diederichs <[log in to unmask]>:
>
>> On Thu, 14 Jul 2016 19:01:37 +0200, Almudena Ponce Salvatierra <
>> [log in to unmask]> wrote:
>> ...
>> >I have no experience at all with twinning and I would like to ask for help
>> >to "quickly" detwin it, I know there are some programs... I tried myself
>> >with XDS, but I couldn't because the data are merohedrally twinned it
>> >seems.
>>
>>
>> don't shoot the messenger! XDS is just a data processing program; it
>> cannot "fix" your data. But it can just process them in the usual way, and
>> you'll have to deal with twinning in the downstream steps (those that come
>> after XDS).
>> Nowadays you might _not_ want to use CCP4 "detwin" program, but rather
>> - try to solve the structure with MR, or experimental phasing
>> - rely on the refinement program to take care of the twinning
>>
>> HTH,
>>
>> Kay
>>
>
>
>
>--
>Almudena Ponce-Salvatierra
>Macromolecular crystallography and Nucleic acid chemistry
>Max Planck Institute for Biophysical Chemistry
>Am Fassberg 11 37077 Göttingen
>Germany
>
|