Saurabh - has does some of the initial investigations of the PaCMAP results
TODOs
Ruge - Separate out upload to s3 PR with just the needed changes so it's not dependent on merging the run from firebase code
Saurabh - Pack peroxisomes in real cell shapes
August 15th, 2023
Update
Got code from Matheus on how to reshape PILR into 3D
plot PILR in 3D
discuss how to compare areas in 3D, start averaging over the phi angle
Need a working group meeting. How many packings per each shape.
Approved 2 open PRs, need a few changes on the upload to s3 PR
TODOs:
Saurabh - Open 1 line PR to fix linting issue: DONE
Ruge - Address changes on upload to s3 PR - still working, merge approved PRs - DONE
Megan - go through the backlog and assign/close/edit issues -DONE
August 1st, 2023
Update
PaCMAP on individual PILRs does not show any trends
Try basic PCA
Waiting for 3D PILR reshaping from Matheus
PR for vectorizing mesh inside-outside points
Compromise between speed and memory usage - make it an option (?)
Try packing peroxisomes in actual cell shapes
have the meshes ready
script is ready to run
waiting to start a packing run for all peroxisome cell shapes
PR to check if PR exists before running workflow is ready for review
Submitted abstract to ASCB
submitted for talk AND poster
TO-DO:
Pack peroxisomes in cell shapes
use variable count?
replicates/randomness
discuss with WG to figure out best next steps
July 18th, 2023
Update
We had two meetings about problems and future directions of correlations analysis.
Suggestions:
PCA on individual PILRs
3D PILR, with progressive subdivisions. Average over phi
Can run multiple packings in parallel on the same recipe with meshes now
In progress: upload results to AWS automatically and open in Simularium browser window
June 20th, 2023
Update:
We are presenting at CellularMorph on Thursday: mostly on correlation analysis for punctuate structures
Review of the database upload/download system
Set up creds system for firebase
TODO:
Include 1 and 3 from below in a slide about our analysis progress
Megan: make problem statement slide for CellMorph WG meeting
Megan: review open PRs
Saurabh: Open 1 more PR
Ruge: new issue, setting up S3 write (could use the same creds setup for AWS credentials)
June 6, 2023
Updates:
Can run PILR analysis on a dockerized image in parallel: example run for 305 peroxisome packing
Can upload gradients to firebase - waiting for review/feedback
Can pack ingredients with variable counts when running multiple packings
TODO:
TIFF image PR comments, refactoring - Saurabh
Parallel packing with meshes
Markdown writer updates - Aditya
Add tests for variable counts, use the same random seed for packing - Saurabh
Run recipes from firebase (waiting for upload PR review/merge) - Ruge
Ideas:
Gradually increase resolution of the radial PILR correlation, starting with 2 bins to separate nuclear vs cytosolic structures, continue to finer bins for radial variation.
May 23, 2023
Updates
TIFF output is done
Ruge waiting on reviews on upload tests
Markdown writer - has a path bug that needs to be fixed
Saurabh has packings running in parallel
TODO:
April 25, 2023
Updates
Megan out until the 10th.
Saurabh has first version of TIFF output - spheres and meshes
Need to review Ruge's PR
Aditya has been working on the markdown writer
TODO:
Saurabh will look into gradient error
Aditya will work on sphere tree implementation
Saurabh will be presenting cellPACK at lunch meeting
April 11, 2023
Updates
Aditya finished updates to analysis PR workflow
Megan and Saurabh combined changes to Env and Analysis code
Small changes: Saurabh fixed ploty plots for multi runs and Megan added recipe info to simularium metadata file header
Ruge added unit tests for upload PR - needs review now
Ruge: separate PR for additional tests, also ready for review
Saurabh has new analysis code, Megan is working on a fix to partner packing schema still
Saurabh PR for live analysis
TODO/Discussion points:
Megan - add Jessica to TIFF meeting
March 28, 2023
Updates:
Presented at WG meeting
Agreed jitter as is does not work - Draft PR has debugging changes that we can make more formalized
Worked on upload duplicate check
Analysis code: able to be run for each ingredient
TODO/discussion Points:
Discuss the possibility of implementing a feature to save voxelized images
Dan, Alli, Jessica
Live analysis mode - options from the analysis config file
Schedule pair programming sessions to work on saving voxeling outputs - Megan
Megan - fix partner packing using keys
Ruge - unit test for upload
Aditya - PR for workflow code
Saurabh - PR for new analysis code
Mar 14, 2023:
Update:
Finished partner packing
Gradient packing new schema is in review
Able to identify first collision in jitter
TODO/discussion points:
discuss save voxelized image/ schedule pair programming?
Present updates at march 23rd meeting
New partner packing: also be able to define a compartment as a partner
combinations of stochastic packing and gradients
Mar 8, 2023:
Updates:
Vector Gradient work
Partner packing (almost) ready for PR
Firebase upload (almost) ready for PR
TODOs:
Tests for Partner Packing - Megan - DONE
Make sure repelling works - DONE
Make "center" optional for vector gradient mode - Saurabh - DONE
Format version migration for new gradient settings -Saurabh (after partners PR) - DONE
Analysis report for partners: get partner distance distribution - Saurabh - in progress
Code cleanup - Ruge - in progress
Mar 2, 2023:
Updates:
Work on GradientData class to validate gradient options
TODOs:
look into build_directional_weight_map and get_direction_length. Refer to new gradient schema wiki for details on directional gradient. - DONE
Jan 31, 2023:
Updates
PR Updates:
Finished the scaled distance PR
Any outstanding issues from Ruge's PR (megan will review today)
Update and continued discussion of jitter error issue - debugging idea to run analysis at each step to identify the first ingr that is packed too close to identify issue
notes from discussion:
are point distances being updated with a -1 when a packing failed at that location?
effects of jitter attempts number, if we increase it a large amount will the error go away?
what are the biases in how we are sampling the jitter space around a grid point? square root of r to avoid more likely sampling at the center of the sphere.
more efficient selection of point within sphere without having to reject the outside of the sphere
debug sparse grid spacing aberration
TODOs:
run similarity analysis on new scaled distance code (Saurabh)
consider pulling packing id from recipe version to display in similarity heatmaps
post images to AWS for auto validation (Megan)
compare contents of local items and database items instead of just strings (Ruge)