Output lamella mask mrc file after void segmentation

[linhua-tai]

Hi Mart,

It was nice seeing you again on GRC! Just a reminder based on our discussion, I think it would be very helpful if you could add in a feature, that in pom contextualization, aside from assigning distance to particles, also output an individual mrc file, for instance, after void segmentation, so that this mrc file can serve as lamella mask during particle extraction of template matching jobs.

Meanwhile, could I ask for another thing? I think it would also be very helpful to data processing if you could add in thickness measurement after void segmentation, not sure if this is an easy thing to do?

Thank you very much for your help!

Best,
Linhua

[mgflast]

Hi Linhua,

I'll include that in a future update. For now, the 'pom contextualize' command can also do exactly that: if you have starfiles from template matching, you can run a command like pom contextualize particles.star --samplers void:0.5:-2 --output particles_context.star and then discard all particles (e.g. using a Relion Select job) for which the sampler returned a negative value. It has the same effect - though I can see how plugging a mask into a pytom_tm job is a little bit more intuitive.

As for thickness measurement: it's easy to do when tomograms are good and the void segmentation accurately annotates the lamella boundaries. But when tomograms aren't great - e.g. totally misaligned, or there is a lot of actual empty area in the tomogram - it is a little bit trickier. This because void mixes i) regions outside of the sample, and ii) poorly aligned tomograms, large bits of ice contamination, super thick super low contrast stuff, basically 'bad data' as defined by my original annotations.
So: I'm working on it, and it's a tool that would fit the easymode toolkit well, but it might be a little while before it release this.

Best,
Mart