zhonge
cryoDRGN-ET force --ntilts
Add option to force --ntilts for cryodrgn backproject_voxel and/or cryodrgn train_vae.
It would be useful to have an option to exclude particles with tilts < --ntilts. For example, if a single tomogram has 14 tilts but every other tomogram has a minimum of 30 tilts. Currently, in this case, --ntilts 30 defaults to 14.
Suggested implementation:
Add --forcentilts to write out something likeforce_ntilts_ind.pkl then subsequently use these indices for cryodrgn backproject_voxel or train_vae. The pkl file should contain indices for all particles that have at least the number of tilts specified by --ntilts.
My preference would be to keep commands that do reconstruction and commands that create indices separate from one another — thus to implement this as part of an e.g. cryodrgn_utils select command that incorporates the functionality of the current cryodrgn_utils select_random command, maybe with this and some other features like --first, --every-n, etc.
If we do go down the above route I think it'd also be a good idea to extend --first, currently only in backproject_voxel, to the other training methods. Happy to hear other suggestions though!
Correct, currently specifying --ntilts X in cryoDRGN-ET takes the min(X, min ntilts in the dataset). I agree with Michal that it probably makes more sense to separately create an index selection to exclude particles that don't have enough tilts as a pre-processing step.
In the past, I've done this separately in a jupyter notebook / Python which I'm happy to share, or we can add this functionality to cryodrgn_utils select.
Add option to force
--ntiltsforcryodrgn backproject_voxeland/orcryodrgn train_vae.It would be useful to have an option to exclude particles with tilts <
--ntilts. For example, if a single tomogram has 14 tilts but every other tomogram has a minimum of 30 tilts. Currently, in this case,--ntilts 30defaults to 14.Suggested implementation: Add
--forcentiltsto write out something likeforce_ntilts_ind.pklthen subsequently use these indices forcryodrgn backproject_voxelortrain_vae. The pkl file should contain indices for all particles that have at least the number of tilts specified by--ntilts.
Hello, it seems like you have successfully run through cryoDRGN-ET's training process. May I refer to your files (particles_from_M.star)? I have already asked the official team for help, but I haven't received a response, so I'm reaching out to you for assistance. Thank you very much.
