Coloring labels by a continuous variable fixed #344
Conversation
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #344 +/- ##
==========================================
+ Coverage 84.27% 84.32% +0.05%
==========================================
Files 8 8
Lines 1558 1557 -1
==========================================
Hits 1313 1313
+ Misses 245 244 -1
|
for more information, see https://pre-commit.ci
…ears-to-be-broken-labels-associated-with-wrong-colors' of github.com:scverse/spatialdata-plot into bugfix/342-coloring-labels-by-a-continuous-variable-appears-to-be-broken-labels-associated-with-wrong-colors
| if np.any(color_source_vector.isna()): | ||
| cell_id[color_source_vector.isna()] = 0 | ||
| val_im: ArrayLike = map_array(seg, cell_id, color_vector.codes + 1) | ||
| val_im: ArrayLike = map_array(seg.copy(), cell_id, color_vector.codes + 1) |
There was a problem hiding this comment.
hmm is the copy really required here? I thought that map array returns an array with same shape as the original array, but it would not have the same pointer
There was a problem hiding this comment.
Ah, I remember. Don't ask me why, but if I don't copy(), one of the tests fails with:
ValueError: buffer source array is read-only
There was a problem hiding this comment.
I had the same issue with the current main: whenever I tried to color labels by a column in table.obs, I'd get ValueError: buffer source array is read-only, and putting in those .copy() calls fixed it. This happened regardless of whether the SpatialData object was buffered to disk or not.
There was a problem hiding this comment.
I see, this is probably a bug in skimage actually. For now we could keep it as it is, but please add a comment with the issue so we remove this when we can.
There was a problem hiding this comment.
preapproving for the vmin vmax fix, but please if you can open an issue in skimage regarding the buffer error. I have seen similar errors in pandas at some point, I believe it had to do with some cython code not allowing for read only buffer, but this could be fixed I think and then prevents a copy.
timtreis
deleted the
bugfix/342-coloring-labels-by-a-continuous-variable-appears-to-be-broken-labels-associated-with-wrong-colors
branch
As reported by @clwgg, there was a bug with the color of continuous annotations which resulted in them having incorrect colors. This is now fixed and the tests have been updated. Furthermore, within this fix I noticed our testing threshold was way too high and let a bunch of bugs pass. So I also lowered that and fixed the bugs this revealed.