Dear Sam,
You first need to know if the low completeness of the inner resolution shell is because of the data collection strategy that you used, or whether reflections have actually been rejected. Poor completeness could be the result of not collecting a sufficiently large rotation range, or a large backstop shadow (although this is a bit unlikely).
You can use the strategy option in iMosflm to determine what the expected completeness is for your rotation range and space group as a function of resolution, and if you mask out the backstop shadow it will also take account of that. You can then compare the expected and achieved completeness. The strategy will also take account of the gaps between the tiles if you are using a tiled detector.
No reflections are normally rejected by iMosflm, but some are flagged (listed as bad reflections in the mosflm.lp file and the number on each image is also plotted when the images are integrated), and by default these are rejected by AIMLESS, but you can choose to accept them as well(see the KEEP keyword). The number of reflections rejected for various reasons is listed in the AIMLESS log file.
Another reason for rejecting reflections in AIMLESS is if the total fraction for summed partials lies outside the range 0.95 to 1.05, if there are a lot of these you could extend this range to (say) 0.8 to 1.1 (keywords PARTIALS TEST). The number rejected as again listed in the AIMLESS log file.
Best wishes,
Andrew
On 8 Jun 2016, at 11:07, Sam Tang <[log in to unmask]> wrote:
> Hello
>
> About a month ago I raised an issue of low innershell completeness in the ccp4bb, and colleagues have given valuable suggestions to this problem. Thanks again for these.
>
> Unfortunately we came into the same issue again in our recent beamtime - innershell completeness of 76% while all other statistics look good. We are quite convinced that there should be some systematic issue with the CCD detector or related settings.
>
> While last time we merged two datasets so that in the end the completeness was (luckily) OK, this time, though, we do not have another set to merge. So we would like to know if we can (1) look at which low resolution reflections were actually rejected by imosflm and during Aimless, and (2) perhaps 'force' the programme to accept these reflections?
>
> Your advice would be highly appreciated.
>
> Kind regards
>
> Sam
>
> Biochemistry Programme, School of Life Sciences, CUHK
|