Re: [ccp4bb]: Scaling Problems

["Eleanor J. Dodson" <ccp4@ysbl.york.ac.uk>]

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

I guess the evidence says: wrong spacegroup.

Checks:
   Do native Patterson to 4A ( see GUI - prepare for MR)
Is there a large off origin peak?

Does it have one coordinate 0.5? If so you cannot tell from absences 
whether that axs is a 21 screw or a 2 fold..

Could there be two molecules related in a somewhat special position? If 
so it may affect you Rmerges- you will get classes of reflections much 
weaker than others..

It also can make it hard to determine the space group.
We had a structure with a NCS translation of 0.03 0 0.5, and after a 
great deal of thought realised this was either P212121 or P21 21 2 . The 
ha sites could be found making either assumption and it was a pain to 
sort out.

The only reason for a high Rmerge in P212121 though if the symmetry is 
truly Pmmm is if there are many weak reflections. Look at the 
diffraction with hklview and see if there are alternating srong and weak 
layers along any axis..

Eleanor

pls3@duke.edu wrote:
> ***  For details on how to be removed from this list visit the  ***
> ***          CCP4 home page http://www.ccp4.ac.uk         ***
> 
> I am currently having difficulty scaling data from a protein-DNA crystal.
> Data was collected previously on this crystal form to about 3.2 A.  It
> processed and scaled as orthorombic with cell dimensions of roughly 39,76,
> and 256  A.  Data was more recently collected on an anomalous edge in
> order to obtain experimental phases, which is where the scaling problems
> have arisen.  I am using HKL2000 to index, integrate, and scale the data
> and all help is welcomed and greatly appreciated.
> 
> Statistics from Scalepack on a previous occasion of the same crystal form
> at a remote wavelength with little or no anomalous signal.
> 
>  Shell Lower Upper Average      Average     Norm. Linear Square
>  limit    Angstrom       I   error   stat. Chi**2  R-fac  R-fac
>       50.00   6.89  4388.1   131.0   114.3  1.885  0.060  0.069
>        6.89   5.47  1113.3    38.6    34.7  1.390  0.082  0.077
>        5.47   4.78   847.5    39.7    37.1  1.116  0.103  0.089
>        4.78   4.34   647.8    41.3    39.6  0.953  0.132  0.109
>        4.34   4.03   610.6    44.4    42.8  0.950  0.149  0.112
>        4.03   3.79   368.0    45.8    45.2  0.818  0.222  0.232
>        3.79   3.60   317.7    50.4    49.9  0.703  0.241  0.212
>        3.60   3.45   454.3    59.6    58.0  0.636  0.164  0.076
>        3.45   3.31   622.2    70.4    68.2  0.547  0.097  0.045
>        3.31   3.20   174.1    72.2    71.9  0.594  0.216  0.132
>   All reflections   1011.6    59.7    56.3  1.020  0.100  0.075
> 
> 
> 
> Spacegroup      P2(1)2(1)2(1)
> Unit Cell       39.821  76.064  256.236
> I/sigma         16.9 (2.4)
> Redundancy      6.5
> Rejections      0.77
> Completeness    97.1% (80.0%)
> 
> Molecular replacement attempts were made and a solution obtained.  One of
> the two complexes in the asymmetric unit showed nice electron density for
> both the model used in molecular replacement and sidechains/residues not
> included in the search.  However, the second protein-DNA complex had only
> moderately continuous density for the core of the search model and nearly
> no additional features outside of the search.
> 
> Recently, data was collected on the Zn anomalous edge in order to obtain
> experimental phases exploiting the bound Zn ions in the protein.  Indexing
> and integration of the data frames was straight-forward and problem free.
> However, attempts to scale the data have proved very problematic.  Below
> is a similar summary of data from Scalepack at the anomalous wavelength.
> The error scale factor for Scalepack was set at 1.3, which is standard and
> the same as used in the first data collection.  Obviously, the chi^2 and
> rejections are incredibly high.  The detector distance and direct beam
> positions used to index the frames is known to be accurate due to other
> data on other projects collected before and after this data set.  Those
> data sets indexed, integrated, and scaled as normal with no problems.
> Additionally, data was collected on this crystal form at a remote
> wavelength and the data similarly doesn't scale well, so I don't believe
> it to be an issue with separation of anomalous pairs.
> 
>      Shell            I/Sigma in resolution shells:
>   Lower Upper      No. of reflections with I / Sigma less than
>   limit limit     0     1     2     3     5    10    20   >20  total
>   50.00  6.24     6    10    21    25    42   124   329  2379   2708
>    6.24  4.96    10    26    40    58   114   299   750  2206   2956
>    4.96  4.33    42    74   120   190   296   604  1311  1713   3024
>    4.33  3.94    30    90   147   222   406   881  1719  1279   2998
>    3.94  3.65    71   166   291   453   761  1424  2379   661   3040
>    3.65  3.44    74   220   471   733  1202  1963  2555   418   2973
>    3.44  3.27   176   509   938  1293  1839  2381  2691   208   2899
>    3.27  3.12   287   917  1536  1900  2293  2594  2805   169   2974
>    3.12  3.00   505  1390  2039  2331  2545  2759  2889    68   2957
>    3.00  2.90   772  1890  2447  2647  2815  2971  3040     5   3045
>  All hkl       1973  5292  8050  9852 12313 16000 20468  9106  29574
> 
> Spacegroup      P2(1)2(1)2(1)
> Unit Cell       39.50     75.70   255.29
> I/sigma         23.9 (1.6)
> Redundancy      3.0 (separating anomalous pairs)
> Rejections      12.56%
> Completeness    94.0% (95.3%)
> 
> Any suggestions on the source and solution to this problem are greatly
> appreciated.
> 
> Paul Shaffer
> Duke University Medical Center
> pls3@duke.edu
> 
>