Re: [ccp4bb]: I to F's

[Bart Hazes <bhazes@ualberta.ca>]

***  For details on how to be removed from this list visit the  ***
***    CCP4 home page http://www.dl.ac.uk/CCP/CCP4/main.html    ***

> Thanks for the answers but I guess we are just in the beginning....
> 
> There is no mention to I(+) and I(-)...?!?!?!

I(+) and I(-) are handled identical to Imean. The only difference is that
during data processing Imean is based on all symmetry related observations
whereas I(+) and I(-) use only the subset for the corresponding Friedel mate.
The standard deviations are also originally calculated by the data processing
scaling/merging step. They include the individual standard deviations of each
observation for a certain unique reflection and generally apply some extra
corrections to make sure that the observed differences between symmetry
related reflections makes sense w.r.t. the standard deviations of the
reflections (I believe the standard deviations normally get inflated
somewhat more).


> I guess truncate will do that but let's suppose a simple case where I do
> not use the TRUNCATE YES keyword. My questions still holds! How standard
> deviations for
> 
> F(+), F(-), and also DANO and SIGDANO are calculated?

I don't remember by heart how TRUNCATE calculates the standard deviations for
amplitudes. For consistency I would think they derive it from the truncated
intensity distribution but you'll have to read the paper if you want to be
sure. SigDano can be calculated directly from the SigF(+) and SigF(-)
reflections as sqrt[SigF(+)**2 + SigF(-)**2]

> How I(+), I(-), SIGI(+) and SIGI(-), which are all that we get from a difraction pattern,
> 
> are combined (in a formula) to generate Imean, SIGIMean, F(+), F(-), DANO
> and SIGDANO ?

See above

> For example; in CNS
> 
> if I >= 0 and sigI < I:
> 
>       F = sqrt(I)
>       sigF = F - sqrt(I - sigI)
> 
> if I >= 0 and sigI >= I:
> 
>       F = sqrt(I)
>       sigF = F
> 
> if I < 0:
> 
>       F = 0
>       sigF = 0

I find this conceptually less attractive than the TRUNCATE approach, but in
practice it may not make much of a difference. It may be more interesting to
consider refinement against intensities which is possible in CNS. In that case
you avoid the whole problem and according to the experts there are some other
advantages to using intensities. However, again, in practice the difference
may be minimal.

> In CNS F(+) and F(-) are treated separetelly and we do not have DANO and SIGDANO.

Chances are that CNS is calculating Dano and SigDano on the fly when it needs
them from F(+) and F(-) using the same calculation to get SigDano as I listed
above. I personally prefer this over the CCP4 method of storing Dano and
SigDano since some information is lost in the conversion (e.g. you cannot
calculate exact values for F(+) and F(-) given Fmean, Dano and SigDano).

> In CCP4 we have DANO and SIGDANO.... So how do we obtain them ?
> 
> I hope this discussion could help us all to understand that !!!

I hope so too.         Bart

> Best regards
> 
> Nagem.
> 
> --
>  ______________________________________________________________________
>  Ronaldo A. P. Nagem                         |
>  Protein Crystallography Group               | E-mail:  nagem@lnls.br
>  LNLS Laboratorio Nacional de Luz Sincrotron | Phone:   55-019-2874520
>  Rua Giuseppe Maximo Scolfaro, 10.000 Guara  | Fax:     55-019-2877110
>  CEP 13084-971 Campinas SP Brasil            | Website: www.lnls.br
> 
> 
> 

===============================================================================

Dept. of Medical Microbiology & Immunology
University of Alberta
1-15 Medical Sciences Building
Edmonton, Alberta, T6G 2H7, Canada
phone:	1-780-492-0042
fax:	1-780-492-7521

===============================================================================