How to speed things up

[shanggutianzhen]

# -------------------------------
sp.tl.spatial_tendency(
    adata,
    mask,
    radius=1000,
    location='out',
    frac=5
)

This step of my code has been running for nearly 24 hours, and it's still not finished, but I only have one Visium data, in contrast to the other Visium, which can be finished very quickly

[shanggutianzhen]

And the previous step

-------------------------------

mask = sp.tl.get_mask_from_domain(adata, clusters='Hipp', KSize=35, cluster_key='region')
WARNING:root:adata has not been initialized, and adata.uns['SOAPy'] has been established
[26352 26662]
And the others are

WARNING:root:adata has not been initialized, and adata.uns['SOAPy'] has been established
[6476 6706] such a range

[wheat-whq]

If the number of genes in the two samples is not significantly different, the contour of the mask can be viewed using some visualization packages(e.g. opencv). The smoother the image boundary is and the fewer the connected components are, the easier it is to calculate. This can be optimized by increasing the 'KSize' parameter in sp.tl.get_mask_from_domain().