Fix Bioimaging inference tool #1580

anuprulez · 2025-02-24T16:21:47Z

Fixes #1544 issue

* Add missing required_files to run remotely Fix the following error: ``` Fatal error: cannot open file '/jetstream2/scratch/main/jobs/65039573/tool_files/scripts/compare.R': No such file or directory ``` * Replace ubuntu-20.04 runner https://github.blog/changelog/2025-01-15-github-actions-ubuntu-20-runner-image-brownout-dates-and-other-breaking-changes/ * Lint R scripts

bgruening · 2025-02-24T16:31:43Z

tools/bioimaging/bioimage_inference.xml

    ]]>
    </command>
    <inputs>
        <param name="input_imaging_model" type="data" format="zip" label="BioImage.IO model" help="Please upload a BioImage.IO model."/>
        <param name="input_image_file" type="data" format="tiff,png" label="Input image" help="Please provide an input image for the analysis."/>
        <param name="input_image_input_size" type="text" label="Size of the input image" help="Provide the size of the input image. See the chosen model's RDF file to find the correct input size. For example: for the BioImage.IO model MitochondriaEMSegmentationBoundaryModel, the input size is 256 x 256 x 32 x 1. Enter the size as 256,256,32,1."/>
+        <param name="input_image_input_axes" type="text" label="Axes of the input image" help="Provide the input axes of the input image. See the chosen model's RDF file to find the correct axes. For example: for the BioImage.IO model MitochondriaEMSegmentationBoundaryModel, the input axes is 'bzyx'"/>


could this be a select?

btw. a text param can also have select options ...

can't it be extracted from the rdf so that the user doesn't need to know this?

The parameter axes provides the real shape (number of dimensions) of the input image/matrix required by the model to make inferences. The parameter input_size is necessary to slice pixels from different input image dimensions. So, both parameters I think are important to ascertain the target size of the image to be provided to the model. RDF file is not available inside the tool from where we could extract the axes information. Also, In a few RDF files, the axes information is not straightforward and written like (from 3d-unet-arabidopsis-apical-stem-cells model):

inputs: - axes: - type: batch - channel_names: - channel0 id: channel type: channel - id: z scale: 1 size: 100 type: space - id: 'y' scale: 1 size: 128 type: space - id: x scale: 1 size: 128 type: space

Users have to infer the correct axes information from above. From above, it is bczyx. Other RDF files have this information directly given as bczyx. (bczyx == batch, channel, z, y, x)

I can provide this as a select option.

@FynnBe I understood that the info in the RDF file is standardised, right?

bgruening · 2025-02-25T09:53:03Z

tools/bioimaging/bioimage_inference.xml

+            <param name="input_image_file" value="input_platynereisemnucleisegmentationboundarymodel.tif" location="https://uk1s3.embassy.ebi.ac.uk/public-datasets/bioimage.io/organized-badger/1/files/sample_input_0.tif"/>
+            <param name="input_image_input_size" value="256,256,32,1"/>
+            <param name="input_image_input_axes" value="bczyx"/>
+            <output name="output_predicted_image" file="output_platynereisemnucleisegmentationboundarymodel.tif" compare="sim_size" delta="100" />


@anuprulez you can use the new image_diff here https://docs.galaxyproject.org/en/latest/dev/schema.html#id117

bgruening · 2025-02-25T10:27:00Z

If we store the RDF for every model, next to the model on our server and provide this as a second input ... do you think we can then support most models?

We could add a conditional:

try to extract model metadata from model RDF file
set metadata on your own

bgruening · 2025-02-25T10:43:28Z

We might be able to read the RDF file with https://github.com/bioimage-io/spec-bioimage-io

conda install -c conda-forge bioimageio.spec

Don't get me wrong, I think we can merge this PR as soon as it is green. Maybe we can add in a next iteration the automatic extraction?

bgruening · 2025-02-25T11:40:23Z

tools/bioimaging/bioimage_inference.xml

            <output name="output_predicted_image_matrix" file="output_nucleisegboundarymodel_matrix.npy" compare="sim_size" delta="100" />
        </test>
        <test>
            <param name="input_imaging_model" value="input_imaging_model.zip" location="https://zenodo.org/api/records/6647674/files/weights-torchscript.pt/content"/>
            <param name="input_image_file" value="input_nucleisegboundarymodel.png"/>
            <param name="input_image_input_size" value="256,256,1,1"/>
-            <output name="output_predicted_image" file="output_nucleisegboundarymodel.tif" compare="sim_size" delta="100" />
+            <param name="input_image_input_axes" value="bcyx"/>
+            <output name="output_predicted_image" file="output_nucleisegboundarymodel.tif" compare="image_diff" />


If those images are too large we can use asserts:

https://github.com/BMCV/galaxy-image-analysis/blob/07aba15c8f753d650fee9cdd455896a77086a615/tools/superdsm/superdsm.xml#L171

Ok, will try that to check if it works with TIF extensions. There are also NPY files which have similar sizes as TIFF files

just check there size, but using assert-size

anuprulez · 2025-02-25T14:21:21Z

@bgruening can you rerun the tests? I am not sure why the first two tool tests fail. They show the following error:

Job in error state.. tool_id: bioimage_inference, exit_code: 1, stderr: Traceback (most recent call last):
  File "/home/runner/work/galaxytools/galaxytools/tools/bioimaging/main.py", line 7, in <module>
    import imageio
ModuleNotFoundError: No module named 'imageio'

If this is the error, the other tests from number 3 to 6 should also fail. Locally, all of them pass.

Thank you!

bgruening · 2025-02-25T16:53:48Z

This looks good to me!

Should we merge this one, so people can try it.

anuprulez · 2025-02-25T17:24:04Z

Yes, I think so.

anuprulez and others added 14 commits February 14, 2025 17:03

update model order

0e73264

update main

3f85240

add support for png

0630717

update tests

04bd671

update tests

9e88305

update methods

2188630

add tests

e293179

fix formatting

6dd9551

Merge branch 'bgruening:master' into fix_imaging_tool

873d3fc

remove bioimaging.core

303f5da

update tool version

ac3631f

add input param information

1fb8991

add end of file line

7a6aabd

bgruening reviewed Feb 24, 2025

View reviewed changes

update output files

36a7819

bgruening reviewed Feb 25, 2025

View reviewed changes

add image_diff for comparing tiff images in tests

fdc5219

bgruening reviewed Feb 25, 2025

View reviewed changes

update tests

8bea904

rerun tests

ddc4d99

Merge branch 'bgruening:master' into fix_imaging_tool

99b577e

bgruening merged commit e08711c into bgruening:master Feb 26, 2025
11 checks passed

anuprulez deleted the fix_imaging_tool branch February 26, 2025 10:49

kostrykin mentioned this pull request Feb 26, 2025

Update documentation for the 'Process image using a BioImage.IO model' Galaxy tool beatrizserrano/galaxy-image-community#26

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix Bioimaging inference tool #1580

Fix Bioimaging inference tool #1580

anuprulez commented Feb 24, 2025

bgruening Feb 24, 2025

bgruening Feb 24, 2025

beatrizserrano Feb 24, 2025

anuprulez Feb 25, 2025

beatrizserrano Feb 25, 2025

bgruening Feb 25, 2025

bgruening commented Feb 25, 2025

bgruening commented Feb 25, 2025

bgruening Feb 25, 2025

anuprulez Feb 25, 2025

bgruening Feb 25, 2025

anuprulez commented Feb 25, 2025 •

edited

Loading

bgruening commented Feb 25, 2025

anuprulez commented Feb 25, 2025

Fix Bioimaging inference tool #1580

Fix Bioimaging inference tool #1580

Conversation

anuprulez commented Feb 24, 2025

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

bgruening commented Feb 25, 2025

bgruening commented Feb 25, 2025

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

anuprulez commented Feb 25, 2025 • edited Loading

bgruening commented Feb 25, 2025

anuprulez commented Feb 25, 2025

anuprulez commented Feb 25, 2025 •

edited

Loading