Analyse data for ph3 experiments.

You will need to have the cp-utils-hmv toolkit to use the notebook:

pip install cp-utils-hmv

0. This assumes an acquisition of data with 4 channels in the slide scanner at 20x.

• Make sure the name of the .czi files is 'Animal-X-Y-Z_Experimental-Procedure_slide-X.czi
• The underscores are very important. Don't have spaces. The last part (e.g. '..._slide-1') is also crucial.

1. Use https://github.com/HernandoMV/czi-extract-slices in Fiji to export slices. (Using channel 4 and 10um/px)

2. Register using ABBA and save transformation field and atlas annotations: https://biop.github.io/ijp-imagetoatlas/registration.html#slices-registration

• Create a folder called 'QuPath' inside the 'Registration' folder just created by the previous script
• Open QuPath, and drag the folder to create a new project. Add the images created by the script, selecting BioFormats-builder. Close QuPath.
• Open ABBA in Fiji and import the QuPath project.
• Flip the axis as the ABBA atlas is the other way around
• Set slice thickness to 1 micrometer!
• Register once with affine, using channel 1 of the atlas (autofluorescence). Set the proper background value!
• Register with spline with 10 landmarks, also correcting background and correct registration
• Export regions to file, and export atlas coordinates to imageJ.
• Save inside the same folder as the original images

3. Use https://github.com/HernandoMV/czi-roisplitter to generate the ROIs, loading the region of interest (e.g. Caudoputamen)

4. Use 'Group_convert_and_enhace.py' in https://github.com/HernandoMV/Fiji_Custom, which transforms the images into 8-bit and normalizes the intensities by channel and by animal

5. Use 'ImageSequence_Downsampler.ijm' in the Fiji_Custom repo to make a copy of both the DARPP-32 channels and the tdTomato (d2) channels in different directories (needed for cellpose). Do not downsample as we are acquiring at 20x.

• Alternatively, just copy both channels to separate folders so you don't have to duplicate data

6. Run cellpose (https://github.com/MouseLand/cellpose) on both directories: e.g. python -m cellpose --dir /home/hernandom/data/Microscopy_Data/Plasticity/PH3_inmuno/Processed_data/PH308/ROIs--Gce_processed--downsized-1_fileend-2.tif/ --save_tif --no_npy --diameter 38 --pretrained_model cyto --chan 0 --use_gpu

7. Run Inmuno_4channels_XXXXXX.cpproj (https://github.com/HernandoMV/CellProfiler_AnalysisPipelines) in CellProfiler. Modify saving paths accordingly.

8. Run the notebook PH3_Analysis.ipynb, for each mouse.

TODO: Create a google colab notebook for testing.