Dear Petr Both questions, on twin law and R-factors, are mainly questions about definitions and standards.
From Online Dictionary of Crystallography: "The twin law is the SET of twin operations mapping two individuals of a twin." In your P 21 21 21 example with a=b, the set contains four operations: 2-fold rotations around [110] and [-110], and rotations around [001] by 90 and -90 degrees. These operations can be written as k,h,-l; -k,-h,-l; -k,h,l; k,-h,l. Four is because there are four operations in 222 point group. All four are equivalent twin operations for this particular twin. Any of them can be taken as a REPRESENTATIVE of the twin law, but it will be incorrect use of terminology to say that one of then IS a twin law. It is valid to use any of them in calculations. Again, the question about R-factors for data from twinned crystal is a question of definition. The standard definitions of R-cryst and R-free are NOT applicable to such data. (Let's leave alone the discussion on whether and when R-factors are useful statistics). This is because the definitions of R-cryst and R-free use the notion of "observed" structure amplitudes. In case of non-twinned crystal, there is French-Wilson procedure (implemented in truncate program), which is a de-facto standard (good or bad - different story) for calculating "observed" structure amplitudes from "observed" intensities. To calculate the "observed" intensities from "twinned" intensities we need yet another standard, e.g. a standard de-twining procedure, which does not exist. Moreover, refinement programs do not use de-twinning, but, other way round, fit calculated "twinned" intensities to "observed" "twinned" intensities. This is robust. But when it goes to R-factors, there is range of options. In Refmac, "twinned" refinement can be conducted against intensities or structure amplitudes, however calculated. The former is in theory more robust, but it uses definitions of R-factors that results in higher values, especially when there are many weak reflections. For bureaucratic purposes, I may suggest the option of refinement against structure amplitudes (obtained e.g. using truncate). That will give lower R?factors. Alternatively, refinement against intensities can be followed by refmac job with zero cycles using structure amplitudes option. Andrey P.S. An important reminder from Garib: output mtz-file from refmac should not be used in subsequent refinements as input. It is only for map calculation and for internal use in model building pipelines. Please always use mtz-file form data reduction program (e.g. aimless) as input for refmac. This comment is especially relevant for "twin refinement". ######################################################################## To unsubscribe from the CCP4BB list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1 This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/
