shelxD

From: Tiruttani Subhramanyam, Udaya Kumar
Date: 6 December 2011 20:16
hi

I have a 2.9 ang selenomet MAD data set integrated in four space groups P4, P422, P222 and C222

i had to integrate in all the four space groups as I was stuck in refining the model in P422

by using shelxD I found solution in P43, P43212 and p212121 but not in C222

the CC all/weak for them are as follows

P43212     52.51 / 34.57
P43           43.12 / 25.73
P212121   48.63 / 31.74

i donot know how many molecules are there in the asymmetric unit but dimerization of the protein  is know from which i can say it could be even number.

but matthews coef shows high probability for heptamer. so i used that solvent content in all the above runs.

another observation  is that at 3.3 ang resolution the anomalous CC % value are 39.2, 35.3, 32.5 and 28.4 for p43212, p43, c2221 and p212121 respectively.

in all cases P422 looks better in terms of SHELXC stats. i am afraid that there is twinning involved.

could anyone suggest me whether its possible to conclude the right space group among them.


With regards
uday


----------
From: George Sheldrick
SHELXD does not use the solvent content. However since you know how many seleniums it found and presumably know how many methionines are in the sequence, you can use that to estimate the number of molecules is the asymmetric unit. Note that an N-terminal selenomethionine is often disordered and so should not be counted, and that there is a twofold axis in P43212 that could be used to form the dimer.

I would start by throwing the P43212 sites into the autotracing version of SHELXE (with and without -i, just in case it is P41212) and seeing if it traces. If one of these two space groups traces appreciably better than the other, then you have solved it. At 2.9A the traces will not be complete but should be good enough for this purpose. If as seems likely you have NCS, you can improve them by using the -n switch.

George

From: Tiruttani Subhramanyam, Udaya Kumar 

Hi

I forgot to mention the ShelxE beta results in the previous mail

P43212 with 15 sites: Mean FOM = 0.669 , Pseudo-free CC = 69.59 %

P43 with 36 sites: Mean FOM = 0.674, Pseudo-free CC = 70.19 %

P212121 with 36 sites:  Mean FOM = 0.689 Pseudo-free CC = 72.08 %

thank you

With regards

uday

-

----------

From: George Sheldrick 

The pseudo-free CC in SHELXE is not a very decisive figure of merit. By far the best way is to run the autotracing (e.g. with -a5) and to look at the CC for the trace against the native data. A value of more than 25% is almost always solved.

George