Spatial LIANA x cell2cell analysis in R

[Qirongmao97]

Thank you for developing the cell2cell pipeline.

I am curious whether it's possible to perform spatial transcriptomics analysis using LIANA x cell2cell in R, given the challenges of interoperability between Python and R for spatial data.

[dbdimitrov]

Hi @Qirongmao97,

What type of spatial experiment would you like to analyse?

Note that the default approach to combine liana and tensor requires cell clusters. So, your application might depend a lot on the dataset in mind.

Daniel

[Qirongmao97]

Hi Daniel,

Thanks for the quick response :)

We have Visium data in different groups, and we also have the annotation for spots in all the samples.

Best,
Qirong

[earmingol]

Hi @Qirongmao97. If you are able to make a grid of your tissue you can assign the spots to each bin in the grid and annotate that assignment, for each spot, in a new metadata column. Then you can use that column as your context and run the analysis as normally done for the R version.

That's what we did for python here: https://ccc-protocols.readthedocs.io/en/latest/notebooks/ccc_python/S4_Spatial-Decomposition.html
Then you can play with the size of the grid (e.g. 10x10, 4x4, etc). The smaller is the grid, the better is the resolution. However you will also force more sparcity per bin.

Also, it's important to set LIANA to return all scores for LR pairs, regardless of their significance. If I remember correctly it's something like return_all_lrs=True.

I hope this helps.