MNN correction on spatial datasets

[sophieRAIBAUD]

Description

...hello
I am struggling with 2 spatial transcriptomics data to compare, but the 2 tissues were spoted perpendicular to each other
I dont know how to pivot the whole anndata (with image + transcripomic data)
any idea ?
Many thanks for this wonderfull package !!!

[giovp]

hi @sophieRAIBAUD thanks for the interest, what do you mean "pivot" ?

[sophieRAIBAUD]

hello
I mean 90° turn in order to see the pictures (and the data) in the same orientation
Many thanks !!!

[giovp]

maybe scipy can help https://docs.scipy.org/doc/scipy/reference/generated/scipy.ndimage.rotate.html ?

[sophieRAIBAUD]

OK thanks !
ill try it for the image but I also will need to rotate the spatial data

[sophieRAIBAUD]

hello Giovanni,
do you think I need to turn the spatial transcriptomic data in order to perform batch correction with MNN on the 2 data than are not spatially aligned ?
Many thanks !
Sophie

[giovp]

you don't need to do so, only thing is to rotate the spatial coordinates in adata.obsm["spatial"] and the image. MNN only operates on gexp counts.

did scipy.rotate work?

[sophieRAIBAUD]

ah OK thanks
So I rotated
-adata.obsm["spatial"]
-adata.uns["spatial"] ["xxx"]["image"][....]
and the results is really strange I lost all the picture (it's completly zoomed on 1 corner)
I'll get back to you with pictures asap
thanks for your help
ps maybe I dont need to rotate the image for to use MNN ???

[giovp]

@sophieRAIBAUD I believe you don't need to rotate the image for the MNN. I'd be curious to see whether the image and coordiantes before rotation are nonetheless aligned?

[sophieRAIBAUD]

Hello Giovanni,
I have performed MNN on my 2 samples following this tutorial:
https://nbisweden.github.io/workshop-scRNAseq/labs/compiled/scanpy/scanpy_03_integration.html
on the PCA, tsne, and umap graphes, it look pretty cool:

but now I would like to perform a louvain clustering on corrected data
but I lost the image in the cdata:

I am not sure I am doing the right thing
Many thanks for your help !
Sophie

[sophieRAIBAUD]

I found the solution with this notebook
https://nbisweden.github.io/workshop-scRNAseq/labs/compiled/scanpy/scanpy_07_spatial.html
but I had to get back the image data:
corrected.uns['spatial'] = adata_raw.uns['spatial']
it works !

[sophieRAIBAUD]

@sophieRAIBAUD I believe you don't need to rotate the image for the MNN. I'd be curious to see whether the image and coordiantes before rotation are nonetheless aligned?

@giovp next, I'll do it with other data that are not aligned at all
thanks

[giovp]

Hi @sophieRAIBAUD

sorruy for late reply, I think the results looks good. could I ask you to post this type of questions in https://discourse.scverse.org/ with squidpy tag? issues are more for code bugs than best practices and usage. Thank you!