Hi Edward
I don't understand, provided the MR program is doing its job, in this
case there are no alternative choices for the space group; all the
alternative space groups are equivalent and should refine equally well
after MR - if they don't there's something badly wrong with your
original data. There's no need to test for alternative space groups
in this case for the simple reason that there aren't any alternatives!
(it's not like orthorhombic where you may need to test for P222, P2221
etc). The space group should have already been determined when you
processed the data and there's no need to change it, assuming of
course the correct space group is not centered monoclinic at all, but
is say P21, or some orthorhombic space group - maybe that is your
problem? The only reason to change it would be if the MR program
doesn't recognise the name because it hasn't been programmed, in which
case you might have to change the name 'C121' or 'C 1 2 1' to 'C2'
using mtzutils. However from what you say the space group was set to
C2 in the processing so there's no need to change it for the MR or
refinement. I don't understand why the MR program is trying
alternative space groups in this case, since it's completely
unnecessary! Shifting the origin only becomes an issue if you want to
compare your structure with a previously determined isomorphous one:
it sounds like this is not your problem (at least not yet!).
Cheers
-- Ian
On Thu, Mar 11, 2010 at 7:40 PM, Edward Snell <[log in to unmask]> wrote:
> Thanks Ian,
>
> It seems as if Balbes does something to the input data (mtz file) looking at the mtzdump. I think it runs reindex (with the 50% loss in completeness). Unfortunately my molecular replacement solution that refines well is one of these origin shifted groups .... With C2 I have 4 choices including the original (basically the ones you mentioned). I've been fortunate and never had to shift origin before and never come across this. Is there an easy way to do this with my data, i.e. what program should I make use of? Again, I apologize if this is a dumb question, it's embarrassing as I've been doing this for more years than I can remember and just hit the brick wall here.
>
> Cheers,
>
> Eddie
>
>
> -----Original Message-----
> From: Ian Tickle [mailto:[log in to unmask]]
> Sent: Thursday, March 11, 2010 2:10 PM
> To: Edward Snell
> Cc: [log in to unmask]
> Subject: Re: [ccp4bb] I 1 21 1 and C1 21 1 in CCP4
>
> Hi Edward
>
> There's no difference between any of the space groups you mention in
> terms of which is right or wrong, they are all equivalent: for the
> purposes of solving & refining the structure it makes absolutely no
> difference which one is chosen. A2, C2 and I2 differ only in the unit
> cell and the indexing. The convention is to choose the one that gives
> beta as close to 90 as possible (but >=90). In all cases b is chosen
> as the unique axis and c >= a. But that is only a convention which
> ensures that if you determine the structure of an isomorphous crystal
> you will always get the same setting, which just makes it easier to
> determine that they are isomorphous. 'C121' (or 'C 2' or 'C 1 2 1')
> and 'I121' (or 'I 2' or 'I 1 2 1') are simply alternative names for
> 'C2' and 'I2'; if the program doesn't accept them it just means the
> other names haven't been programmed. I've never understood the
> rationale for the longer names, I think it was supposed to make the
> names unique, but provided you stick with the conventional settings
> (and in the vast majority of cases there's no reason not to), the
> short names are already unique without having the extra '1's and
> spaces. It seems to me that having the option of adding '1' & spaces
> just adds to the confusion and leads to the exact problem you have
> encountered! Finally, 'C21' or 'C1211' or 'I21' or 'I1211' (or the
> versions with embedded spaces) are just origin-shifted
> non-conventional settings of C2 and I2; again it's very unlikely you
> would have a good reason to use them.
>
> Personally I would just stick with C2 or I2!
>
> Cheers
>
> -- Ian
>
> On Thu, Mar 11, 2010 at 6:07 PM, Edward Snell <[log in to unmask]> wrote:
>> Dear All,
>>
>>
>>
>> I may be asking a dumb question and if so I apologize. I have a ~200 amino
>> acid N-terminal 'arm' of a full protein (C-term already solved) that
>> diffracts to ~2.1A. It integrates nicely in C2 and gives good molecular
>> replacement models from Balbes in C121, 1I21, A121, C1211 and I1211 (>90%
>> certain). Phaser gives nice hits in C2 and I2 but I have not been able to
>> get it to accept the C1211 or I1211. It appears from refining the data that
>> my solution is either C1211 or more probably I1211 unfortunately I am having
>> problems using these two groups. In addition to this I also have some
>> evidence of twinning L Molrep and Refmac will accept the I1211 and C1211
>> groups but the freer and uniquify components after Scala will not. I've used
>> reindex with the space group as is, in brackets, and in the 'I 1 21 1'
>> notation all to no avail. My refinements have been forced to use the mtz
>> files output by the Balbes server which loose 50% completeness in these two
>> cases. My current R and Rfree with this data are 27% and 38% and the overall
>> chain that I have built is in good agreement with one of the SAXS envelopes
>> I have, but the data completeness issue is causing me problems. Has anyone
>> come across this, do I need to add a symm group to the existing groups? I
>> seem to remember some issues with these a few years back but that part of my
>> brain is too dusty at present!
>>
>>
>>
>> Thanks for any help and hopefully I will not be hitting my head against the
>> wall with the obvious.
>>
>>
>>
>> Cheers,
>>
>> Eddie
>>
>>
>>
>> Edward Snell Ph.D.
>> Assistant Prof. Department of Structural Biology, SUNY Buffalo,
>> Hauptman-Woodward Medical Research Institute
>> 700 Ellicott Street, Buffalo, NY 14203-1102
>> Phone: (716) 898 8631 Fax: (716) 898 8660
>>
>> Skype: eddie.snell Email: [log in to unmask]
>>
>> Telepathy: 42.2 GHz
>>
>> Heisenberg was probably here!
>>
>>
>
|