MAINT: TODO for open PRs

[adamwitmer]

This issue is for logging/tracking TODO items for open PRs migrated from the gitlab instance of the same repo. This list is non-exhaustive based on an initial overview of the current state of the repo and should be expanded to include items such as data, manuscript, etc:

Major items:

• clean up commit history; several PRs have multiple shared/overlapping commits (i.e. ENH: Nidhin opencv detection #2, ENH: Add BubbleSAM detection method #3, ENH: Add scripts to analyze BubbleSAM and OpenCV detection results #4, ENH: Add ML training and inference #5)
• very large PRs (can be split off into smaller PRs?)
• updating the README to describe how to recreate all manuscript figures
• setup .github/ci.yml

Minor items

• needs requirements.txt file for installing required packages

[tylerjereddy]

A few thoughts here:

clean up commit history; several PR's have multiple shared/overlapping commits

That's not too surprising--I had very limited bandwidth to review, so many PRs have accumulated that build on each other/share commits. I'm not sure there is really much action needed on that apart from the usual git workflow--reviewing the PRs one at a time, convincing me they are ready, getting them merged, and then rebasing the next PR in line.

very large PR's (can be split off into smaller MR's?)

Maybe, but the main thing I need help with is getting the code reviewed/tested/fixed up, confirming it matches some of the plots in the manuscript/SI, and that we can also run it on the "raw" data stored on glycan. And possibly a few more PRs to address the review comments on the manuscript proper, like EER thresholding of phase separation determinations.

updating the README to describe how to recreate all manuscript figures

I'm pretty sure Nidhin was doing that to some degree as a response to some of my review comments--probably useful to point out the specific cases where a new capability was added without a corresponding README update. Really getting into the PRs and providing constructive feedback on them directly is probably the most useful thing for the team.

setup .github/ci.yml

Let me work on that today--I see that there is no CI running on any of the 5 open PRs, so that's important to remedy. In the internal repo at https://lisdi-git.lanl.gov/ldrd_dr_neat/ldrd_neat_ml, I see that we have unit tests running with Python 3.12 and a "functional test" that keeps timing out after running for an hour. We should probably not port over the "functional test" since that is basically broken at the moment.

The last time the CI was flushed (2 months ago at https://lisdi-git.lanl.gov/ldrd_dr_neat/ldrd_neat_ml/-/jobs/179835), the pytest tests ran with output: 19 passed, 15 warnings in 6.80s. Replicating that here, perhaps with a few Python versions spread across Linux/MacOS would be a good start. Let me give it a shot.