Re: [ccp4bb] iMosflm twist large variance and specific space group scale output

[Xiao Lei]

Dr. Harry,

Thank you very much for the illustration!

On Wed, Feb 15, 2017 at 2:07 PM, Harry Powell 
wrote:

> Hi
>
> First things first...
>
> 2. I am not sure if iMosflm can be set to output the scale of specific
> space group. For example, in my data set even though the integration is
> done at P321 space group, after click "quick symm" or "quick scale" it
> seems the program automatically choose what it thinks the most likely space
> group P6222, not P321, I know in HKL2000 there is an option to choose
> specific group and scale it, how to do this in iMosflm? I'd like to scale
> the data to P321 and some subgroups in P3 but not P6222.
>
>
> Yes, easily done. In the Indexing task, at the bottom left where iMosflm
> chooses the space group for you, there's a drop-down where you can pick any
> of the chiral trigonal or hexagonal space groups. You could even, if you
> wanted, type in any of the achiral space groups and that would work too.
>
> HOWEVER, unless you are using the Strategy option, I really wouldn't
> bother doing this (most of the time). The *measured* reflections for both
> trigonal and hexagonal space groups are the in the same places and the
> extra symmetry you would have from hexagonal is ignored in the integration.
> The QuickScale option will let Pointless do its job and give you a better
> analysis of the true symmetry and also apply the symmetry to your MTZ file
> (Pointless will use the intensities in its calculations, which is not
> something that Mosflm does).
>
> Importing the output MTZ file from iMosflm (if you haven't run QuickScale)
> into either ccp4i or ccp4i2 will also run Pointless and try to sort out
> your symmetry for you.
>
> Does  this mean if I see twist variance greater than "1" then I need to
> fix both "tilt' and "twist" value before integration in imosflm?
>
>
> Yes, generally speaking if you fix tilt you would probably want to fix
> twist as well.
>
> In my case the variance is greater than 0.5 but less than 1, do I still
> need to fix both "tilt" and "twist" value
>
>
> Maybe, probably. It's hard to be definitive without seeing the graphs from
> integration or the full log file output.
>
> Try it and see if the merging stats look better...
>
> Harry
> --
> Dr Harry Powell
> Chairman of International Union of Crystallography Commission on
> Crystallographic Computing
> Chairman of European Crystallographic Association SIG9 (Crystallographic
> Computing)
>
> On 15 Feb 2017, at 16:36, Xiao Lei  wrote:
>
> Dear CCP4bb members,
>
> I have two questions about imosflm integration and scaling process (my
> data is 3A resolution and 0.5 degree mosaicity):
>
> 1. I have warnings after integration saying the twist variance is too
> large (from -0.26 to 0.33, which is larger than 0.5), suggesting that cell
> parameters may be wrong, and reindex to the right space group may be
> needed. After I read the imosflm tutorials from internet, the tutorials
> says "For weak images the tilt and twist may not be well defined, and might
> refine to large values (greater than one degree). In such cases these
> parameters should be fixed. It is not generally recommended to fix any
> other detector parameters."
>
> Does  this mean if I see twist variance greater than "1" then I need to
> fix both "tilt' and "twist" value before integration in imosflm?   In my
> case the variance is greater than 0.5 but less than 1, do I still need to
> fix both "tilt" and "twist" value?
>
> 2. I am not sure if iMosflm can be set to output the scale of specific
> space group. For example, in my data set even though the integration is
> done at P321 space group, after click "quick symm" or "quick scale" it
> seems the program automatically choose what it thinks the most likely space
> group P6222, not P321, I know in HKL2000 there is an option to choose
> specific group and scale it, how to do this in iMosflm? I'd like to scale
> the data to P321 and some subgroups in P3 but not P6222.
>
>
> Thanks for any inputs.
>
>
>

Re: [ccp4bb] iMosflm twist large variance and specific space group scale output

[Harry Powell]

Hi

First things first...

> 2. I am not sure if iMosflm can be set to output the scale of specific space 
> group. For example, in my data set even though the integration is done at 
> P321 space group, after click "quick symm" or "quick scale" it seems the 
> program automatically choose what it thinks the most likely space group 
> P6222, not P321, I know in HKL2000 there is an option to choose specific 
> group and scale it, how to do this in iMosflm? I'd like to scale the data to 
> P321 and some subgroups in P3 but not P6222.

Yes, easily done. In the Indexing task, at the bottom left where iMosflm 
chooses the space group for you, there's a drop-down where you can pick any of 
the chiral trigonal or hexagonal space groups. You could even, if you wanted, 
type in any of the achiral space groups and that would work too. 

HOWEVER, unless you are using the Strategy option, I really wouldn't bother 
doing this (most of the time). The measured reflections for both trigonal and 
hexagonal space groups are the in the same places and the extra symmetry you 
would have from hexagonal is ignored in the integration. The QuickScale option 
will let Pointless do its job and give you a better analysis of the true 
symmetry and also apply the symmetry to your MTZ file (Pointless will use the 
intensities in its calculations, which is not something that Mosflm does).

Importing the output MTZ file from iMosflm (if you haven't run QuickScale) into 
either ccp4i or ccp4i2 will also run Pointless and try to sort out your 
symmetry for you.

> Does  this mean if I see twist variance greater than "1" then I need to fix 
> both "tilt' and "twist" value before integration in imosflm?

Yes, generally speaking if you fix tilt you would probably want to fix twist as 
well.

> In my case the variance is greater than 0.5 but less than 1, do I still need 
> to fix both "tilt" and "twist" value

Maybe, probably. It's hard to be definitive without seeing the graphs from 
integration or the full log file output.

Try it and see if the merging stats look better...

Harry
--
Dr Harry Powell
Chairman of International Union of Crystallography Commission on 
Crystallographic Computing
Chairman of European Crystallographic Association SIG9 (Crystallographic 
Computing) 

> On 15 Feb 2017, at 16:36, Xiao Lei  wrote:
> 
> Dear CCP4bb members,
> 
> I have two questions about imosflm integration and scaling process (my data 
> is 3A resolution and 0.5 degree mosaicity):
> 
> 1. I have warnings after integration saying the twist variance is too large 
> (from -0.26 to 0.33, which is larger than 0.5), suggesting that cell 
> parameters may be wrong, and reindex to the right space group may be needed. 
> After I read the imosflm tutorials from internet, the tutorials says "For 
> weak images the tilt and twist may not be well defined, and might refine to 
> large values (greater than one degree). In such cases these parameters should 
> be fixed. It is not generally recommended to fix any other detector 
> parameters." 
> 
> Does  this mean if I see twist variance greater than "1" then I need to fix 
> both "tilt' and "twist" value before integration in imosflm?   In my case the 
> variance is greater than 0.5 but less than 1, do I still need to fix both 
> "tilt" and "twist" value?
> 
> 2. I am not sure if iMosflm can be set to output the scale of specific space 
> group. For example, in my data set even though the integration is done at 
> P321 space group, after click "quick symm" or "quick scale" it seems the 
> program automatically choose what it thinks the most likely space group 
> P6222, not P321, I know in HKL2000 there is an option to choose specific 
> group and scale it, how to do this in iMosflm? I'd like to scale the data to 
> P321 and some subgroups in P3 but not P6222.
> 
> 
> Thanks for any inputs.
> 
> 

[ccp4bb] iMosflm twist large variance and specific space group scale output

[Xiao Lei]

Dear CCP4bb members,

I have two questions about imosflm integration and scaling process (my data
is 3A resolution and 0.5 degree mosaicity):

1. I have warnings after integration saying the twist variance is too large
(from -0.26 to 0.33, which is larger than 0.5), suggesting that cell
parameters may be wrong, and reindex to the right space group may be
needed. After I read the imosflm tutorials from internet, the tutorials
says "For weak images the tilt and twist may not be well defined, and might
refine to large values (greater than one degree). In such cases these
parameters should be fixed. It is not generally recommended to fix any
other detector parameters."

Does this mean if I see twist variance greater than "1" then I need to fix
both "tilt' and "twist" value before integration in imosflm?   In my case
the variance is greater than 0.5 but less than 1, do I still need to fix
both "tilt" and "twist" value?

2. I am not sure if iMosflm can be set to output the scale of specific
space group. For example, in my data set even though the integration is
done at P321 space group, after click "quick symm" or "quick scale" it
seems the program automatically choose what it thinks the most likely space
group P6222, not P321, I know in HKL2000 there is an option to choose
specific group and scale it, how to do this in iMosflm? I'd like to scale
the data to P321 and some subgroups in P3 but not P6222.


Thanks for any inputs.