Yes, as the twinning fraction increases from 0 to 0.5, the cumulative intensity distribution curve changes in a continuous way from untwinned to perfectly twinned. The exact way in which it does this was calculated by Rees (Acta A 36, 578 (1980)). Note that the variation is markedly non-linear - if the plot is 'a little off' the untwinned plot the twinning fraction may be rather more than you think, whereas if the plot is 'a little off' the perfectly twinned plot, the twinning fraction will still be very close to 0.5.
The usual caveats apply - for example the shape of the cumulative intensity distribution can be affected by pseudosymmetry as well as twinning, so while the cumulative intensity distribution can indeed be used to detect partial as well as perfect twinning, it is important to consider other measures (e.g. moments) as well. Norman -----Original Message----- From: CCP4 bulletin board [mailto:[EMAIL PROTECTED] On Behalf Of Bryan W. Lepore Sent: 30 October 2007 17:29 To: CCP4BB@JISCMAIL.AC.UK Subject: Re: [ccp4bb] Pseudo-merohedral twinning and Molecular replacement On Mon, 29 Oct 2007, Iain Kerr wrote: > The cumulative intensity distribution plot from crystal A did suggest > partial twinning (attached, doesn't look too bad though..) notwithstanding other plots/statistics, does the cum. intens. dist. plot (e.g. from truncate) really show a continuum from untwinned to twinned? i.e, if the plots are 'overlapped in the middle', no question - twinned. but, if the plots are 'a little off, but not in the middle' can this result (alone) really mean the data is - as we want to say - partially twinned? i.e. is the plot robust only for the detection of perfect twinning? -bryan
