Analyze Visium H&E data with h5ad

[asmlgkj]

Description

due to the server network, I can not run img = sq.datasets.visium_hne_image() adata = sq.datasets.visium_hne_adata()

so I downloaded the visium_fluo.h5ad on another computer and uploaded it to the server. I tried to split the data into image and adata, but

import pandas as pd import os import matplotlib.pyplot as plt

import anndata as ad import scanpy as sc import squidpy as sq

sc.logging.print_header() print(f"squidpy=={sq.version}") adata = sc.read_h5ad('visium_fluo.h5ad') img = adata.uns['spatial']['V1_Adult_Mouse_Brain_Coronal_Section_2']['images']['hires']

but the following code failed img.show(channelwise=True) ttributeError: 'numpy.ndarray' object has no attribute 'show'

how can I reproduce the tutorial on the downlaoded data

thanks a lot ...

Minimal reproducible example

...
scanpy==1.9.5 anndata==0.9.2 umap==0.5.6 numpy==1.24.3 scipy==1.11.3 pandas==2.2.2 scikit-learn==1.3.0 statsmodels==0.14.1 igraph==0.11.5 pynndescent==0.5.12
squidpy==1.5.0

## Traceback

<!-- If applicable, replace `...` with an error traceback below: -->

<details>

```pytb
...

Version

...

[asmlgkj]

Thaks a lot. I found it hard for a new learner to use the downloaded data, can there be a new tutorial start from the downloaded data, not the inner function.

tutotial lin is https://squidpy.readthedocs.io/en/stable/notebooks/tutorials/tutorial_visium_hne.html

import pandas as pd import os import matplotlib.pyplot as plt

import anndata as ad import scanpy as sc import squidpy as sq import tifffile as tiff sc.logging.print_header() adata = ad.read_h5ad('visium_hne.h5ad') img = tiff.imread('visium_hne.tiff') library_id = list(adata.uns['spatial'])[0] adata.uns['spatial'][library_id]['images'] = {'hires': img}

sq.pl.spatial_scatter(adata, color="cluster")

img_ml = sq.im.ImageContainer(img) for scale in [1.0, 2.0]: feature_name = f"features_summary_scale{scale}" sq.im.calculate_image_features( adata, img_ml, features="summary", key_added=feature_name, n_jobs=4, scale=scale, ) adata.obsm["features"] = pd.concat( [adata.obsm[f] for f in adata.obsm.keys() if "features_summary" in f], axis="columns", ) adata.obsm["features"].columns = ad.utils.make_index_unique( adata.obsm["features"].columns )

helper function returning a clustering

def cluster_features(features: pd.DataFrame, like=None) -> pd.Series: """ Calculate leiden clustering of features.

Specify filter of features using `like`.
"""
# filter features
if like is not None:
    features = features.filter(like=like)
# create temporary adata to calculate the clustering
adata = ad.AnnData(features)
# important - feature values are not scaled, so need to scale them before PCA
sc.pp.scale(adata)
# calculate leiden clustering
sc.pp.pca(adata, n_comps=min(10, features.shape[1] - 1))
sc.pp.neighbors(adata)
sc.tl.leiden(adata)

return adata.obs["leiden"]

calculate feature clusters

adata.obs["features_cluster"] = cluster_features(adata.obsm["features"], like="summary")

compare feature and gene clusters

sq.pl.spatial_scatter(adata, color=["features_cluster", "cluster"])

this generate a figure without cluster information, I do not the reason

[LLehner]

Thaks a lot. I found it hard for a new learner to use the downloaded data, can there be a new tutorial start from the downloaded data, not the inner function.

The functions

img = sq.datasets.visium_hne_image() adata = sq.datasets.visium_hne_adata()

just download the data, so there is no need for a different tutorial. However if you work with visium data I would recommend you to use a specified reader from spatialdata-io to do that, which will return a spatialdata object, which makes working with spatial omics data a lot more efficient.

Regarding your other comments, i'm not quite sure what your question is.

[asmlgkj]

Thanks a lot. @LLehner Because of the network problem in China, download the data is not successful. My question is I read data from the pre-downloaded data, and the cluser does not appear on the function sq.pl.spatial_scatter(adata, color=["features_cluster", "cluster"]) the right result is in the link https://squidpy.readthedocs.io/en/stable/_images/c24920a65ab8f67b572b44646959e7f2f078ce7cc19f7147e5a07ff93451525b.png

[giovp]

hi @asmlgkj , I think the cluster are computed using the notebook tutorial, so if you run the tutorial it should work!