SWC to Tiff conversion for some files results in OverflowError

[n-shamsi]

When using the cli tool to convert a .swc to .tiff, there is an overflow error resulting in skeleton.py from osteoid

command used: kimimaro swc to --format tiff /path/to/.swc

relevant system info:
python==3.12.9
kimimaro==5.2.0
microviewer==1.13.1
numpy==2.2.5
osteoid==0.3.2
macOS Sequoia 15.4.1

result:

  File "/Users/nish/miniconda3/envs/test/bin/kimimaro", line 8, in <module>
    sys.exit(main())
             ^^^^^^

  File "/Users/nish/miniconda3/envs/test/lib/python3.12/site-packages/click/core.py", line 1442, in __call__
    return self.main(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/Users/nish/miniconda3/envs/test/lib/python3.12/site-packages/click/core.py", line 1363, in main
    rv = self.invoke(ctx)
         ^^^^^^^^^^^^^^^^

  File "/Users/nish/miniconda3/envs/test/lib/python3.12/site-packages/click/core.py", line 1830, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/Users/nish/miniconda3/envs/test/lib/python3.12/site-packages/click/core.py", line 1830, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/Users/nish/miniconda3/envs/test/lib/python3.12/site-packages/click/core.py", line 1226, in invoke
    return ctx.invoke(self.callback, **ctx.params)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/Users/nish/miniconda3/envs/test/lib/python3.12/site-packages/click/core.py", line 794, in invoke
    return callback(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/Users/nish/miniconda3/envs/test/lib/python3.12/site-packages/kimimaro_cli/__init__.py", line 152, in to_image
    skel = Skeleton.from_swc(f.read())
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/Users/nish/miniconda3/envs/test/lib/python3.12/site-packages/osteoid/skeleton.py", line 1157, in from_swc
    return Skeleton(vertices, edges, radii, vertex_types)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/Users/nish/miniconda3/envs/test/lib/python3.12/site-packages/osteoid/skeleton.py", line 104, in __init__
    self.vertex_types = np.array(vertex_types, dtype=np.uint8)
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

OverflowError: Python integer -79 out of bounds for uint8```

[william-silversmith]

Hi! Thanks for the bug report! Generally speaking, the vertex types are supposed to be positive integers in the specification I've been following. Is the vertex type actually -79 in your SWC?

[n-shamsi]

The dataset I am interested in has both v3dpbd.swc and .swc files, and I can use either one of those to make a tiff stack. So, I am providing unaltered inputs (swc files) to the kimimaro cli utility (attached a sample, in case it helps with recreating the error). I am looking into this myself, and if I find a fix which works, I'll open a pull request, in osteoid or kimimaro_cli depending on the fix. Thank you for looking into it! But no urgency on fixes

Archive.zip

[william-silversmith]

Thank you! I took a look at the SWC files and some of the vertex types were actually negative. Are these files meaningful and valid? One possible fix is to change uint8 into int8, though that could have unintended consequences for some users. Usually the upper range of the uint8 isn't used so much, but I don't know how people actually use vertex types in the wild.

[william-silversmith]

Just to add on, because if the negative values are not meaningful to you, just open the SWC file in a text editor and remove the - signs from the second column.

[n-shamsi]

From what I can tell, negative values for vertex can be meaningful, and it depends on how someone annotated the data, http://www.neuronland.org/NLMorphologyConverter/MorphologyFormats/SWC/Spec.html.

But whether the negative vertex types are included in the data from the Skeleton class, or turned positive using np.abs or sth, doesn't affect the tiff output from the swc file, unless someone wants to convert from such a swc of course :)

My sample of data is from here, and I am just using the data as provided. I am okay closing the PR in osteoid, and this issue, if there are possible unintended downstream consequences. Up to you ^^