Running Analysis using our own Data

[amilworks]

Use Case

We would like to use the entire workflow you built on our own data. For instance, is this the first step to run when using our own images?

make_cellprofiler_image_set \
    --image_csv ./quilt_data_contrasted_test/metadata.csv \
    --defaults_json fish_morphology_code/cellprofiler/cellprofiler_image_set_defaults.json \
    --path_key rescaled_2D_fov_tiff_path \
    --local_path ./quilt_data_contrasted_test \
    --out_loc ./test_image_set_list.csv

The metadata.csv that was downloaded automatically from Quilt when I retrained the model looked like this

Unnamed: 0	DataPath	AnnotationDiffusePath	AnnotationFibersPath	AnnotationDisorganizedPunctaPath	AnnotationOrganizedPunctaPath	AnnotationOrganizedZDisks
0	DataPath/9e529b4f_data.tif	AnnotationDiffusePath/6d19e326_diffuse_other.gz	AnnotationFibersPath/f5733b1e_fibers.gz	AnnotationDisorganizedPunctaPath/58b75b7c_disorganized_puncta.gz	AnnotationOrganizedPunctaPath/3f479a6e_organized_puncta.gz	AnnotationOrganizedZDisks/6b0cb327_organized_z_disks.gz

Solution

Can you provide some documentation, just a simple README would suffice, on the Steps required to run the entire workflow using our own data? In particular, I notice there are several CSV files like metadata.csv and others in the Inference section that I am unsure whether we create or use the ones created from your analysis.

We would like to get to the point where we can retrain the model using our own data, but would like to replicate the data preprocessing you did to get to that point of training, if that makes sense.

Thanks for all the great work on this paper!

[ayushi-agrawal-gladstone]

I am interested in this as well. @amilworks Were you able to get this workflow working for your images?

[TadeasFrycak]

I am interested in this as well. @amilworks, @ayushi-agrawal-gladstone were you able to get this workflow working for your images?