search

fmi-basel / gliberal-scMultipleX

Feature extraction and linking of multiplexing data.
BSD 3-Clause "New" or "Revised" License
8 stars 0 forks source link

Add automated testing for Fractal measurements & masked measurements #81

Closed jluethi closed 1 year ago

jluethi commented 1 year ago

I've been running into a measurement bug in Fractal and thus decided it was time to build some automated testing. Here's a draft for it.

This PR doesn't actually close my original issue (https://github.com/fmi-basel/gliberal-scMultipleX/issues/79) yet, because I haven't yet been able to reproduce it in the tests. But I think it's valuable in any case that we'd add automated testing.

These tests here would add the following:

  1. OME-Zarr example data (tiny OME-Zarrs) with measurement tables as scMultipleX generates them at the moment
  2. Tests that run measurements on 2D & 3D, with or without morphology on this dataset and compare it with the expected measurements contained in the OME-Zarr file
  3. A github automation that should trigger these tests automatically upon pull requests & commits to main.

I'm not fully sure whether the github automation is already working it, I'm opening the PR to test it.

jluethi commented 1 year ago

Hmm, running into a weird error now that does not occur locally. Had some issues with uploading the example data to github though, so not sure whether something could have gotten corrupted. I'll need to investigate further

jluethi commented 1 year ago

This PR now also includes support to make measurements in masked ROIs, i.e. measure all the cells within an organoid, but don't make measurements of other cells in the bounding box (but not in the organoid mask).

Also, adding it to the FMI Fractal setup now and testing whether it solves #79 . In local tests, #79 was running fine now.

jluethi commented 1 year ago

https://github.com/fmi-basel/gliberal-scMultipleX/pull/81/commits/7ca9cea9096f23ebacf0bebe24a5bab09b467b40 closes #79

jluethi commented 1 year ago

Hey @nrepina This PR fixes the issues I originally hit, as well as adding quite extensive automated testing for the Fractal scMultipleX task to ensure that we get the same results going forward :) From my side, it's ready to be merged (and already deployed at FMI)

nrepina commented 1 year ago

This looks great, thanks @jluethi ! Merging now and will test it next week on the scmultiplex side.