Hi Lisa,
The problem you have can be very tricky. Here my comments:
-What do you mean by "I can not solve the structure by molecular replacement"?.
Moleculare replacement programs usually produce some solution, even if it is
wrong. Do the R-factors stay around 50%, is the electron density very bad? If
you do not get a solution at all, the solution might be rejected because of
clashes. In this case you may want to cut out some variable loops from the
model or increase the maximum number of allowed clashes.
-What does the Matthews number tell you? If you get a very dense crystal
packing even with one molecule in the asymmetric unit, you most likely have
twinning.
-3.3 Å means rather poor data. This means that the statistics you get on your
space group and on the twinning will be tentative and you have to test all
possibilities. Also you may have pseudo crystallographic NCS making the
apparent space group look different from the true space group. This means that
in molecular replacement, you have to test all possibilities, e.g. P622, P6122,
P6222, P6322 etc. Most molecular replacement programs have an option to
automatically test them all.
-Considering twinning, you have to process the data in the lower symmetry space
group, presumably P6 and submit the data to a twin server, as Fred suggested. I
would also run molecular replacement on the lower symmetry data. Even with
perfect twinning, one usuallly gets two solutions, corresponding to the two
twin domains.
Good luck!
Herman
________________________________
From: CCP4 bulletin board [mailto:[email protected]] On Behalf Of
LISA
Sent: Tuesday, January 22, 2013 7:25 AM
To: [email protected]
Subject: [ccp4bb] perfect twin
Hi all,
Does anyone know how to check a data is twin and how to detwin?
My data is 3.3 A with space group P6222 or P6422? But I can not solve
this structue by molecular replacement with the model has 42% sequence
identity. I am wondering my data maybe twin.
Thanks
Lisa
