OME-TIFF Read / Write

[srivarra]

• added OME-TIFF support

If you haven't already, please read through our contributing guidelines before opening your PR

What is the purpose of this PR?

Closes #807. Adds support for converting a FOV to an OME-TIFF and vice versa.

How did you implement your changes

Added two functions, fov_to_ome and ome_to_fov in load_utils.py.( moved to tmi)

• fov_to_ome: Can convert multiple FOVs with whatever channels necessary to OME-TIFFs
• ome_to_fov: Can only convert one OME-TIFF to a FOV.
• Added a Jupyter Notebook which walks users through the conversion process.
• Added an OME-TIFF into the Example Dataset.

Remaining issues

• Merge HF PR 10
• Merge General Updates to sync with ark-analysis changes alpineer#16 (moves the ome_tiff relevant code to tmi)
• Merge Adding TMI #825

[ngreenwald]

Looks good

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[srivarra]

@cliu72 I've added a short utility notebook which walks a user through the conversion process. Let me know your thoughts!

[cliu72]

This looks good to me! I think it would be good to include in the notebook description that the conversion is to single-channel TIFs, instead of ambiguously saying it's in a format for ark. Maybe something like: "This notebook is used to convert an OME-TIFF into single-channel TIFFs (one TIFF per marker), which is the format used throughout Ark-Analysis."

Also, I think it would be good to have a more descriptive title for the notebook - like "OME-TIFF_Converter" or something like that?

Is there also going to be a notebook going the other way (i.e. single channel -> OME), or is that going to be a task for later? (I only need this one for my paper so I don't really care lol)

[srivarra]

@cliu72 I fixed the issue for reading the second image. You need the imagecodecs library in order to decompress the first image (the second image requires it as well).

imagecodecs does not have a native Apple Silicon PYPI build, that architecture's version is only available through Anaconda, but for any other platform it should be pip installable.

Also the Example Dataset does not include any OME-TIFFS, would you like a few of those in for this converter notebook in particular?

[cliu72]

If it's not too much work, can we add an example OME-TIFF? I think you can just take one of the example FOVs, use this notebook to create the OME-TIFF, and then add it to the example dataset. I think just one FOV would be fine.

And then can we move the cell to download the example dataset to the relevant sections? I.e. move the get_example_dataset for single channel TIFFs to the section "Single Channel TIFFs to OME-TIFF", and then add the cell to get the example OME-TIFF to the first section. Since presumably most users will be only doing one of the two.

[srivarra]

@cliu72 Dataset is available on this branch, Need to fix a couple of tests, but you can use it with the conversion notebook.

[cliu72]

Sweet, looks good!

Last thing, can we add more description to before the cell to download the example dataset? Maybe so it looks like this:

OME-TIFF to Single Channel TIFFs

If you would like to test this feature using an example dataset, run the cell below. To use your own data, skip the cell below and change ome_dir_path and data_dir_path below to point to your own data.

example_dataset.get_example_dataset(dataset="ome_tiff", save_dir = pathlib.Path("../data/example_dataset/"), overwrite_existing = True)

[rest of the code]

Single Channel TIFFs to OME-TIFF

If you would like to test this feature using an example dataset, run the cell below. To use your own data, skip the cell below and change data_dir and save_dir below to point to your own data.

[rest of the code]

[cliu72]

Not sure if there are remaining merge/example dataset issues to resolve, but I think this looks good to me

[ngreenwald]

There are some merge conflicts that need to be addressed. It will also change the requirements, which will require a new docker. We'll discuss at our meeting today if it makes sense to get all that done today, or wait till after the closure

[srivarra]

@ngreenwald Missing some coverage, as I'm not able to create ome-tiffs with tifffile that have a slightly different format (i.e. a couple of the example ones that @cliu72 tested out).

[ngreenwald]

Code looks good, but I think we should hold off on adding TMI and making other changes until we've resolved any issues with the current python build. We haven't put out a new release yet that has a compatible docker image for the current version, we should do that and iron out any other kinks. Then we can make additional releases with these other PRs that will further change the requirements.

[ngreenwald]

Looks good!