Hi,
A common minor annoyance I run into is that if for whatever reason exposure groups have been lost (e.g. after RBMC), but a particle set was refined with beam tilt groups, CS will complain that there is variation of a CTF param within an exposure group (necessitating reassigning particles to micrographs or otherwise regenerating the previous exposure groups, which can be a hassle if that was done many jobs ago).
But if the original particle set has already been refined with beam tilt groups, it ought to be possible (and sometimes easier) to infer the grouping based on the variation of the CTF parameters. Would it be possible to add an option to e.g. Exposure group utilities to split the particle set into exposure groups based on e.g. variation in refined beam tilt?
Cheers
Oli
1 Like
Hi @olibclarke,
Thanks for the report. We wanted to clarify what you’re experiencing when you see the exposure groups being lost, e.g. after RBMC — what exactly have you noticed in this regard? Once exposure groups are assigned, they should remain part of the particles and exposure datasets indefinitely, in the sense that the assignment persists in the dataset and can only change if reassigned, but definitely let us know if you’re seeing contrary behaviour.
In the case that CTF values vary within an exposure group (e.g. caused by separate Refinement or CTF refinement jobs on two particle stacks from the same exposure group), re-combining these and re-refining CTF will cause issues. The easiest way to workaround this is to use Exposure Group Utilities in combine&set mode, inputting all of the particles from one exposure group, and set the “Combine strategy” to take mode, which will set all high-order aberrations to be constant across the exposure group (equal to the most common value across the dataset). These output particles can then be used for CTF refinement.
The inverse process definitely makes sense as a feature in Exposure Group Utilities – to split/reassign particles into exposure groups based on the refined CTF values, such that each output exp group has constant high-order aberrations within it – I’ll add this to our tracker 
Best,
Michael
2 Likes
not sure exactly how it happened - will need to track back - but I know I have encountered it several times and wished I had the inverse process available, rather than having to dig out or regenerate the xml files to redo the groupings (after which I use the take mode you mention)
1 Like