Hi Kevin,

I would say that the most useful place for this token will be in
reduced-data land prior to getting phases - so for instance going into
Phaser it could make a sensible choice about what spacegroups to test
based on the accuracy information, likewise experimental phasing / dm
programs. Once these programs have computed phases they will (most
likely) be in a position to decide the correct spacegroup and write this
out with some authority along with the phases. 

So what I am getting at is that you are unlikely to have phases without
knowing what the right spacegroup is. Certainly, you are unlikely to
have correct phases ;o)

This cannot possibly "cure" the spacegroup naming problem, but it could
certainly allow people to say what they mean when assigning some
spacegroup information to a reflection file... 

Cheers,

Graeme 

-----Original Message-----
From: CCP4 bulletin board [mailto:[EMAIL PROTECTED] On Behalf Of
Kevin Cowtan
Sent: 09 November 2007 14:12
To: [email protected]
Subject: Re: [ccp4bb] pointless (1.2.0) and enantiomorphic SG's

This is great! But before we all get carried away, a quick reality check
is in order. This isn't an area I've thought about, but my limited
understanding raises the following issues....

Given that an MTZ file contains a cell, an indexing of the data, and
possibly phases involving a choice of origin, the use of an MTZ file
immediately implies that some decision has been made about the
spacegroup.

Selecting a different spacegroup from a list of possibilities may
involve reindexing, and/or changing origin/enantiomorph dependent
properties such as phases and HLs.

It would be possible for an MTZ file to say something like this:
  'This file is generated assuming spacegroup X. Spacegroups X1, X2, X3,
X4, X5.... are also possible.'.

It should also be possible to compute on the fly whether a particular
column type needs to be modified for a different spacegroup.

How will we present this in user interfaces? As a starting point, until
a spacegroup has been settled upon, the user will need the option of
picking the spacegroup after having selected the MTZ. This means
potentially a reindexing step added to every single task - messy. As an
intermediate, being able to change the 'currently selected' spacegroup
as a separate task would be an improvement.


Phil Evans wrote:
> I agree
> 
> On 9 Nov 2007, at 13:50, Eleanor Dodson wrote:
> 
>> As I often say!!! The mtz format should carry point group and 
>> alternate SGs - then be upgraded when youknow the correct SG..
>>
>> Eleanor
>>
>> Phil Evans wrote:
>>> It just picks the first in the list, to store in the output MTZ 
>>> file, which can only handle one (92 < 96) Phil
>>>
>>>
>>> On 8 Nov 2007, at 22:26, Bryan W. Lepore wrote:
>>>
>>>> when pointless (1.2.0) finds enantiomorphic SG's, what is the 
>>>> criterion for 'Selecting' one over the other?
>>>>
>>>> e.g. i ran pointless on some tetragonal data, and the enantiomorphs

>>>> SG92/SG96 are selected as strong candidates
>>>>
>>>>    Spacegroup         TotProb SysAbsProb     Reindex
Conditions
>>>>    <P 41 21 2> ( 92)    0.956  0.956                         00l: 
>>>> l=4n, h00: h=2n (zones 1,2)
>>>>    <P 43 21 2> ( 96)    0.956  0.956                         00l: 
>>>> l=4n, h00: h=2n (zones 1,2)
>>>>
>>>> ... then pointless reports :
>>>>
>>>> Selecting space group P 41 21 2 as solutions are enantiomorphic
>>>> Best Solution     space group P 41 21 2
>>>>
>>>> ... is that b/c pointless can only report one, and SG92 came up
first?
>>>>
>>>> -bryan
>>>
>>>
>>
> 
> 

Reply via email to