It would be nice to be able to flip particle coordinates in Y within cryosparc when importing particles (instead of flipping micrographs in Y during extraction).
When extracting particle picked outside of cryosparc with the “flip micrographs in Y” option there are issues downstream: when inspecting particle picks, the mics and particle locations do not match as coordinates are still Y-flipped (and there is no option here to flip mics in Y). More importantly, I presume that subsequent local refinement would also use the wrong particle locations.
My current workaround is to import particles to cryosparc, then use csparc2star.py to create a star file of Y-flipped particles and then re-import these into cryosparc.
What is your current workflow where this becomes an issue? Are you performing motion correction and picking outside of CS? Are you moving between CS and external software multiple times while processing a dataset?
In response to your current workaround, why not use csparc2star.py to flip picks prior to import into CS? Additionally, would it be possible to import you mics, flip your initial particle stack with cs2star.py, import those particles into CS, and then associate those particles to micrographs such that you can always re-extract in CS along with performing local motion correction/RBMC?
Thanks for following up - I have some small and very low SNR particles where CS Patch Motion correction appears to perform quite poorly. I ended up using MotionCor3 for Motion and CTF correction outside of CS, which significantly improved subsequent topaz denoising (enabling seeing particles by eye) and also improved picking (deep learning based picking, as well as template/blob picking inside CS), as well as initial 2D/3D classification.
The csparc2star.py workaround (while annoying) works fine - but it appears counter-intuitive to me to only have the option to temporarily flip micrographs during particle extraction in CS, but not having the options to:
a) permanently flip micrographs upon import,
b) flip particle coordinates (e.g. in particle tool set?) or
c) choosing to not flip raw movies upon import (which appears to be the root cause for all this confusion).
Please let me know if any of these options are actually available and I missed them!
Thanks for detailing your workflow. As a suggestion, it may possible for you to import your externally processed micrographs (in .mrc format) using the Import micrographs job. Additionally, particle stacks can be imported (in .star format) using Import particle stack and as an input, the imported micrographs can be supplied as an input so that the particles are linked to the micrographs. Then, any additional downstream processing steps should work in CryoSPARC without additional Y-flips.
Just as a background note, the reason flipping in the Y-axis is sometimes necessary is that several cryo-EM software packages implement different conventions for the order of pixels in .tiff files. CryoSPARC reads them in the order they are stored on disk, while some other packages apply a flip as well. Thus there may be a mismatch when importing particle locations from an externally processed dataset and trying to use micrographs preprocessed in CS.
The “Import movies” job applies a Y-flip (raw movies in .tiff format), while the “Import micrographs” job (external motion corrected images in .mrc format) does not apply a Y-flip (or vice versa). So this creates a headache when combining particle stacks picked on either set of images
Having an option to permanently flip micrographs and/or particle coordinates within Cryosparc would be quite useful in my opinion.