WG meetings 2024 - mesoscope/cellpack GitHub Wiki
July 23th, 2024
Updates and agenda:
- Lysosomes are not punctate - not worth focusing on in initial analysis
- Multiple gradient PRs merged
- Markdown Class PR!
- Firebase supports gradient arrays!
- cellPACK server work: got access to AWS Batch (can run jobs) and ECR (can push images)
- Dockerized version of cellPACK is running
- Setup hydra for cellpack analysis pipeline
- Github workflows and linting for cellpack analysis
- ASCB abstract drafted
- New packings with peroxisomes and endosomes with count obtained from observed data
TODO:
- Saurabh: Review gradient arrays in Firebase PR
- Ruge and Alli: Working on getting recipes to pack using image on ECR
- Saurabh: Run punctate analysis on new packings
- Ruge: Troubleshoot MarkdownWriter class github workflow
June 25th, 2024
Updates and agenda:
- PR open for grid_file_path upload
- endosome analysis is done
- pair programming on mixed/combined gradients
TODO:
- review PR
- in progress: uploading refactor
- discuss lysosomes (in SciSim meeting)
June 11th, 2024
Updates and agenda
- PR for loading grid files from s3: need get it to work with firebase upload
- SAC is over
- Discussed grid file paths
TODO:
- upload grid file path to firebase on upload
- need a second approval on open PR
- revisit upload algorithm (debug)
- refactor cellpack analysis repo, and do packing with more variation: a more faithful version of the distribution
May 28th, 2024
Updates and agenda
- the slides for SAC are done
- plots for cellpack results in simularium file is working
- two prs for cleanup
- https links work for loading meshes from s3
- merged github action cache so the actions run faster
TODO (next meeting):
- discuss feasibility of uploading a whole bunch of recipes
- discuss grid file paths in firebase?
May 14th, 2024
Updates and agenda
- SAC writeup and slides
- Loading meshes from s3 does not work - will need to update
MeshStore.get_mesh_filepath_and_extension(filename)
- github action cache - discuss upgrading to pyproject.toml and python 3.10
- discuss cleaning up unused recipes and config options from repo
- if time: discuss questions from the previous server chat
Notes
- Use https:// link instead of s3 URIs for loading meshes from s3
Action Items:
- Saurabh: loading grid files from s3
- Ruge: add plots to result files
- Saurabh: check priority packing (removed functions included some priority calculations) and other functions moved to utils.py. Get rid of unused code completely and move older analyses to the cellpack-analysis repository.
- Saurabh: test pdm build backend to manage dependencies, will need to update test workflows as well
- Megan: do a first pass of the cellPACK 1 page summary
- Saurabh: Incorporate edits from Megan and Graham into cellPACK 1 page summary
April 30th, 2024
Updates
- Presented at cell morph WGF meeting on 25th April
- Saurabh was able to upload and run a peroxisome recipe through firebase
- Lunch meeting presentation prep
TODO:
- Saurabh: test loading meshes from s3
- Saurabh: loading grid files from s3
- Saurabh: look at Ruge's PR
April 16th, 2024
Updates
- Completed: Normalize obtained nuclear distance distribution with the available space to see if we recover a flat line
- Progress in analysis pipeline
- Question around using firebase:
- first need to upload local recipes using the
upload
script - also possible to run a json from AWS? worth pursuing?
- first need to upload local recipes using the
- Question around using firebase:
- Ruge has open PR for skipping db system in the case that a user is only doing local packing
- last presentation April 9: available space investigations
- Proposal for cellPACK server (talk about more today?)
TODO:
- for normalized distribution: 300 cells data pooled together. Try doing it per cell and then plot ratio
- ideas: two theories for the observed plot ratio: is it just excluded space? Ie the ER is keeping the peroxisomes from being close to the nucleus? or do they actually have a preference for the edge of ER
April 2nd, 2024
Updates
- Discussion on grid point selection for biased packings
TODO:
-
compare distance distributions across elongation shape mode bins
-
Normalize obtained nuclear distance distribution with the available space to see if we recover a flat line
a. Can also be used as a test for deviation from randomness
-
Use z position as a metric to look at endosome distribution
March 19th, 2024
Updates
- Set up meeting with Alli re: cellPACK server on 20th March
- Merged PR 224, 237
- Firebase packing path updates
- Published a new version on PyPI
- removed cvapipe analysis dependency from the cellpack-analysis repo
- Presented updates at the scientific simulations WG - punctate analysis on observed peroxisome data
TODO:
- Saurabh: review PR #230
- Saurabh/Karthik: Install cellPACK from PyPI in the cellpack-analysis repository
- Lunch meeting updates on April 17th - Follow up and outline on 2nd March
- Ruge: issue #238
March 5th, 2024
Updates
- Presented at BPS
- Started cellpack infrastructure meetings (pipeline optimization/tooling)
- generated synthetic data for representation learning group
TODO:
- talk to Alli (working with Ruge to make a cellpack server)
Feb 2nd, 2024
Updates
- removed panda3d (in review)
- Met with Jessica to discuss pipeline optimization/tooling, going to have an ongoing meeting to coordinate efforts
TODO:
- Path validation PR to open
- PR comment
- see if we can take a cellMorph meeting after BPS to have Julie see new data
Jan 26th, 2024
Updates
- fixed issue with surface packing
- Reviewed open gradient image PR (still needs second review, and requested one small change)
- Started reading staging firebase setup PR, but need to finish review
TODO:
- remove panda3d
- start on some issues towards making cellpack usable by other people
Jan 19th, 2024
Update
- we work on #213 but didn't finish, some meshes seem to not be loading correctly
- Worked on exploring the baseline/null hypothesis of packing many replicates and running the analysis on the set to see what we can define as "same" in our system.
- In review: staging firebase setup
TODO:
- some known speed ups (#219)
- PR review (saving gradient data as image)
- #212: firebase validation and deleting metadata