Hi,
Is there any possibility to trace particles from the selected 2D classes back to the original micrographs and see how they look there? Is there something similar to “Manually curate exposures” job, but for particles?
Best,
Alesia
You can use inspect particle picks, with a Select 2D input. A curate particles job has been requested, but does not yet exist
The Manually Curate Exposures job also accepts the particles as an input and can display them, which might be useful depending on what you want to do.
1 Like
Great! Thank you for suggestions. I will try both ways.
1 Like
@olibclarke and @Boszlacs are correct, it is semi automated (select 2D manually => curate exposures).
I generally find myself in this situation to guess if rare views or damaged particles in 2D are coming from different ice. With Rebalance 2D Classes I use 8-16 super classes (split the output =on, so each super class has its own particle stack). I can then use manually curate exposures to see if damaged particles are coming from good micrographs (air-water interface) and the rare views are from micrographs with a particular features.
The “proof” in this case is to run that manual curate exposure job with particles from 1 or more “good” 2D super class and micrographs. Assess how many micrographs are taken, can even run an extract job with 4-8x bin or so. At this point should have all of the micrographs containing “good” particles that align well in 2D. With these micrographs/exposures, a final extract job with rare or damaged particles can determine if they are coming from these micrographs (e.g. what % of the input particles are actually extracted). It is unknown if rare views and damaged particles come from similar or different micrographs.
I’ve used manual picker (also accepts particles as input), to inspect by a 2D class/superclass basis. In several cases with R2/2 UltraFoil, I found one type of 2D class coming from from particles far from the hole edge and a different 2D class from particles close to the hole edge. *you can also manually delete particles.
Does seem like a lot of work if not needed. At least for running TOPAZ training on rare views I believe this is is helpful for accurately estimating particles per-micrograph and the micrographs to use.
If needed for the project, I also use the micrograph IDs to associate a grid square from the metadata.