[ccp4bb]: Question about I to F conversion

[Jeff Taylor <jstaylor@duke.edu>]

***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***

(Sorry if this is a duplicate message)

Hi -

I need some help understanding what I'm seeing in my data completeness
before and after conversion from intensities to amplitudes.  

My data set is from a crystal with a large unit cell and therefore has a
large number of reflections.  I have processed the data with denzo and
scalepack.  To get an idea of the data quality and completeness I looked
at the statistics from a scalepack run using a zero sigma cutoff and
found an overall completeness of 89.3% (see table below).  I then ran
scalepack again using a -3 sigma cut and converted the resulting
reflection file to amplitudes using the CCP4 program Truncate, with
Wilson scaling.  Now when I look at the completeness of the resulting
reflection file in CNS I get an overall completeness of 97.8% (see table
below).

How can it be that there are more reflections with F/sigma(F)>2 than
there were with I/sigma(I)>0 ?

Does Wilson scaling or any other correction applied to the data by
Truncate change the I/sigma(I) ratio for a particular reflection?  Did I
do something wrong in the conversion process?

Thanks for any help,

Jeff Taylor


Scalepack output from a run using a zero sigma cutoff:

     Shell            I/Sigma in resolution shells:
  Lower Upper      % of reflections with I / Sigma less than
  limit limit     0     1     2     3     5    10    20   >20  total
  50.00  5.38   0.0   0.3   1.2   2.1   3.8   7.6  17.3  80.4   97.7
   5.38  4.27   0.0   0.7   1.9   3.2   5.7  11.3  26.2  71.4   97.6
   4.27  3.73   0.2   1.6   4.0   6.7  11.7  22.4  47.8  45.7   93.5
   3.73  3.39   0.3   2.6   6.3   9.9  16.7  32.4  65.5  25.0   90.6
   3.39  3.15   0.3   4.5  13.3  21.7  33.7  53.1  75.1  15.3   90.5
   3.15  2.96   0.4   6.2  19.7  32.5  50.2  70.1  85.3   5.3   90.7
   2.96  2.82   0.3   7.8  25.7  41.7  59.9  77.8  86.4   2.2   88.6
   2.82  2.69   0.5  10.8  33.2  51.3  67.6  79.4  83.5   0.8   84.3
   2.69  2.59   0.5  12.3  37.4  56.5  71.4  79.2  81.1   0.3   81.3
   2.59  2.50   0.6  14.8  43.3  61.6  72.4  76.8  77.6   0.1   77.7
 All hkl        0.3   6.1  18.5  28.5  39.1  50.7  64.3  25.0   89.3



CNS output using the -3 sigma cut scalepack file (after conversion with
Truncate):

 STATistics>            completeness 
 STATistics>        selection=(50 >= d >= 2.5 and amplitude(fobs) > 2
*sigma ) 
 Total of    171175 structure factor elements were selected.
 STATistics>   end 
 #bin | resolution range | #refl | 
    1   5.38   50.00      17668       .9801
    2   4.27    5.38      17451       .9913
    3   3.73    4.27      17260       .9842
    4   3.39    3.73      17196       .9826
    5   3.15    3.39      17080       .9805
    6   2.96    3.15      17055       .9809
    7   2.82    2.96      17066       .9781
    8   2.69    2.82      16940       .9744
    9   2.59    2.69      16804       .9678
   10   2.50    2.59      16655       .9628

    1   2.50   50.00     171175       .9783

-- 
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Dr. Jeffrey S. Taylor       Duke University Medical Center
Department of Biochemistry  Box 3711, Nanaline Duke Bld  
Phone (919) 681-5266        Durham, NC 27710