Reading in MERSCOPE sample data with squidpy.read.vizgen() doesn't work for every dataset?

[Laura-Waechter]

Dear all,
I tried to follow this workflow (https://squidpy.readthedocs.io/en/stable/external_tutorials/tutorial_vizgen_mouse_liver.html) for the Human Liver cancer dataset (https://info.vizgen.com/ffpe-showcase) instead. But when using squidpy.read.vizgen(), the obsm and var from the AnnData object are only filled with NaN, despite the documents containing the same information as the mouse liver dataset. Did anyone have this problem before and knows how to solve it?
Thanks very much in advance!

[giovp]

hi @Laura-Waechter ,

could you try to install form main? pip install git+https://github.com/scverse/squidpy@main

let me know if it works,

[Laura-Waechter]

Thanks for your fast reply @giovp . I now tried to install it the way you suggested, but the matrix is still empty...

[giovp]

can you try with #679 ?

[Laura-Waechter]

I replaced the code of the function with this:
coords = coords.astype("str")
coords.index = coords.index.astype(adata.obs.index.dtype)
but that still doesn't help

[dfhannum]

This issue should be resolved now.

[giovp]

it is in main, thank you again @dfhannum . Am trying to cut release asap but encountering few hiccups, but should manage to do it this week.

[ghar1821]

Hi guys, I'm encountering the same issue.

The read_vizgen function in master threw me the following error:

'DataFrameView' object has no attribute '_view_args'

from this line of code in read function (line 168):
adata.obs.drop(columns=["center_x", "center_y"], inplace=True)

Using the version in conda, I managed to run the function, but then the fov, volume, and center obs data is NaN.

And I'm using the Human Breast Cancer data that is publicly available on vizgen website.

[parkerstevenson]

Hi all, I'm still encountering this issue using squidpy.read.vizgen(), the obs from the AnnData object are only filled with NaN after changing line 164 of the squidpy.read.vizgen() function from coords.set_index(coords.index.astype("str"), inplace=True) to coords.set_index(coords.index.astype(adata.obs.index.dtype), inplace=True)

[giovp]

hi both, I think the version in conda is not up to date yet, so please use the pip installable version