-
Notifications
You must be signed in to change notification settings - Fork 26
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Deepcell upload loop cleaning #1023
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Jus a couple of small things.
For the Retry
call at line 245 num_retries
is not defined.
This PR was actually meant to scrap the retry a certain amount of times logic and instead implement recall run_deepcell_direct() in a loop until success or timeout. Looks like I missed some code that needed to be removed. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
A few additional comments.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Couple of control-flow comments
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks super clean!
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good, just one side question.
Next Release v0.6.4 (#1012) * 0.6.4 * updated environment.yml * Update README.md * updated windows setup docs * added repo verison in git clone * file fix Deepcell upload loop cleaning (#1023) * loop check for successful deepcell upload * update test zip paths * helper functions pls * zip input, upload, and extract per batch * cleaning * docstring indent * fov -> fovs file name * remove overwrite warning * new dir for each test call * fix bad logic warning testing * remove arg from docstring * add helper function test * batch_num start at 1 * add previously processed warning * set timeouts and scrap retry * correct timeout errors and update tests * timeout 5 mins * add to unprocessed list after loop closes * continue loop after extraction * break instead of continue * skipped processing print fixes * adjust print statements * 3 seconds before second zip call update
* plot_cluster Next Release v0.6.4 (#1012) * 0.6.4 * updated environment.yml * Update README.md * updated windows setup docs * added repo verison in git clone * file fix Deepcell upload loop cleaning (#1023) * loop check for successful deepcell upload * update test zip paths * helper functions pls * zip input, upload, and extract per batch * cleaning * docstring indent * fov -> fovs file name * remove overwrite warning * new dir for each test call * fix bad logic warning testing * remove arg from docstring * add helper function test * batch_num start at 1 * add previously processed warning * set timeouts and scrap retry * correct timeout errors and update tests * timeout 5 mins * add to unprocessed list after loop closes * continue loop after extraction * break instead of continue * skipped processing print fixes * adjust print statements * 3 seconds before second zip call update * updated readthedocs python version * updated plot_utils docstring * docstrings v2 * use GS [1,1] if cbar_visible is False, otherwise use GS [1, 2] for plot + colorbar * added cbar_visible T/F for tests * pycodestyle * added numpy list to cmap / norm function * Remove cells that don't have any pixel clusters expressed prior to Pixie (#1051) * Drop cells without any SOM cluster expressions * Add testing for dropping non-expressed cells * Include `img_sub_folder` as parameter in Mantis calls (#1050) * Ensure mantis can read the image sub folder specified * Test on example dataset * Default img_sub_folder to None in create_mantis_dir (standard used elsewhere) * Doc fix and updated example dataset * Update segmentation labels again * Clarify img_sub_folder (mostly for refreshing tests) * Force .github/get_example_dataset.py to use updated revision * Update the file names to look for * Update HuggingFace version (#1053) * v0.1.10 (#1057) * Pin to scikit-image to v0.19.3 (#1060) * pin to 0.19.3 * requirement less than 0.19.3 * Make sure overwrite functionality for pixel clustering propagates into helper functions (#1058) * Propagate overwrite functionality for pixel clustering * Ensure re-normalization prevented on overwrite for pixel SOM re-assignment * normalize_data should be the opposite of overwrite * Single mantis directory (#1061) * single mantis dir, no recopying * post clustering mask suffix * generic clustering adjustments * mask prefix * remove mask prefix * Mass plotting notebook (#1052) * added plotting notebook * updated readmed * git merge issues * git merge fixes v2 - docs/landing.md * type * Remove cells that don't have any pixel clusters expressed prior to Pixie (#1051) * Drop cells without any SOM cluster expressions * Add testing for dropping non-expressed cells * Include `img_sub_folder` as parameter in Mantis calls (#1050) * Ensure mantis can read the image sub folder specified * Test on example dataset * Default img_sub_folder to None in create_mantis_dir (standard used elsewhere) * Doc fix and updated example dataset * Update segmentation labels again * Clarify img_sub_folder (mostly for refreshing tests) * Force .github/get_example_dataset.py to use updated revision * Update the file names to look for * fixed clipped plots in the notebooks. * Update HuggingFace version (#1053) * v0.1.10 (#1057) * Added continuous variable segmentation plots * Pin to scikit-image to v0.19.3 (#1060) * pin to 0.19.3 * requirement less than 0.19.3 * Make sure overwrite functionality for pixel clustering propagates into helper functions (#1058) * Propagate overwrite functionality for pixel clustering * Ensure re-normalization prevented on overwrite for pixel SOM re-assignment * normalize_data should be the opposite of overwrite * Single mantis directory (#1061) * single mantis dir, no recopying * post clustering mask suffix * generic clustering adjustments * mask prefix * remove mask prefix * strange merge issue * removed branch filters on CI, so if you branch off of a branch and run GHA, the CI workflow should work * changed ci.yml, filter push on main branch only * python 3.9 doesn't support | for Union type overloading * added fovs parameter for continuous variable plots * updated notebook to include plots * updated cell clustering notebooks to add erosion (default to True), cleaned up some notebook outputs * fixed style imports * re-added test_generate_summary_stats. Was accidently deleted in a refactoring step. * newline at eof fiber_segmentation_test.py * seaborn-paper -> seaborn-v0_8-paper * added cell to view all plotting styles i mask gen nb * pycodestyle --------- Co-authored-by: alex-l-kong <31424707+alex-l-kong@users.noreply.github.com> Co-authored-by: camisowers <38049893+camisowers@users.noreply.github.com> * default ero * added erode=True for generate_cluster_mask --------- Co-authored-by: alex-l-kong <31424707+alex-l-kong@users.noreply.github.com> Co-authored-by: camisowers <38049893+camisowers@users.noreply.github.com>
If you haven't already, please read through our contributing guidelines before opening your PR
What is the purpose of this PR?
Closes #628. Changes how we attempt to re-upload failed fov batches to the deepcell server.
Also refactors
create_deepcell_output()
with cleanlier helper functions. The code will now also check for previously processed batches, to avoid re-zipping and re-uploading already segmented batches to deepcell.How did you implement your changes
Currently, we attempt to re-upload a failed fov batches until success or 5 retry attempts are reached. A cleaner, more long term fix is to attempt to process each batch once, check if any batch .zip files are missing, and re-run the failed batches.
The re-running continues until either a successful output or the function times out.
fovs_batch_1.zip
exists, if not generate the input zip filedeepcell_output_fovs_batch_1.zip
exists, if not upload input file to deepcell and extract segmentation tiffsRemaining issues