Lachlan Cranswick wrote:
>> Hi All,
>> I am trying to use the OMIT instruction (OMIT 1 70) for Shelxs97 to
> put a limit on ESDs and 2-theta angle. This is for some data
> extracted from powder diffraction data using the LeBail
> method and Fullprof). (looking at the ESDs, the quality of the data
> is not like that of a single crystal) The overall aim is to get
> a Patterson using the FMAP 1 instruction.
>> Problem is that while Shelxs97 seems to say it accepts the OMIT
> instruction, it does not seem to be applying it(?) as we get
> the same result if the OMIT is excluded.
....
=====================================================================
Dear Lachlan
It is usually more efficient to send technical queries about SHELX
directly to the author (as recommended in the manual and homepage!).
However since your question has implications for macromolecular users I
will reply to the newsgroup.
The OMIT instruction in SHELXS-97 flags reflections so that they are not
phased by direct methods, but they are always used for statistical
purposes and for Pattersons. I will try to make this clearer in the
manual next time I revise it.
SHELXS-97 is frequently used for locating heavy atoms from SIR or SAS
delta-F data by automated Patterson interpretation. These data are
often barely significant and applying (say) a 4-sigma(F) threshold (OMIT
4) would often result in most of the data being rejected! Since a
Patterson theoretically requires complete data this may not be the best
approach. In addition the sigma(delta-F) estimates may not be
particularly reliable in such cases.
I did some tests using anomalous delta-F data that I had processed
carefully myself paying special attention to getting good sigma
estimates. It is clearly easier to make reasonable sigma estimates from
SAS data from only one crystal than from SIR when datasets from more
than one crystal have to be combined. I anticipated that weighting down
rather than rejecting the reflections with poor |delta-F|/sigma(delta-F)
ratios would improve the map, but - judged by the signal to noise of the
Harker peaks - all the schemes I tried were worse than simply using unit
weights! What did however improve the Pattersons were schemes to reject
or 'distribution fit' reflections with |delta-F| larger that expected
from the rms value of delta-F in the corresponding resolution shell.
Most SHELXS users use the CCP4 package or (for SAS data) the XPREP
program from Bruker AXS (nee Siemens AXS) to process their data in this
way.
Anyway, it should now be clear why I defined the OMIT instruction in
SHELXS-97 in the way I did. In SHELXL-97 OMIT rejects reflections
completely.
Best wishes, George Sheldrick
