directionality of proximity analysis

[jmintch]

Hi @marcovarrone, I've been using you're tool for a while now and quite like it so far! I've recently started using the nhood_enrichment and diff_nhood_enrichment functions more for my analyses. I understand that the comparisons are directional and therefore the heatmap not symmetrical. I am a bit stuck though with regards to which direction the comparisons are going.

For example, let's take the plot from https://cellcharter.readthedocs.io/en/latest/notebooks/codex_mouse_spleen.html where you ran

adata_balbc = adata[adata.obs['condition'] == 'BALBc']
cc.gr.nhood_enrichment(
adata_balbc,
cluster_key='spatial_cluster',
)

cc.pl.nhood_enrichment(
adata_balbc,
cluster_key='spatial_cluster',
annotate=True,
vmin=-1,
vmax=1,
figsize=(3,3),
fontsize=5,
)

There's one entry with the value 0.73 in row 1 (cluster 0) column 5 (cluster 4). Would this mark the enrichment of cluster 4 around cluster 0, or cluster 0 around cluster 4? Judging by the tissue image I'd guess the first option, but I'm not super sure.

I'd really appreciate a clarification - really just want to make sure I get this right.

Thanks a lot!

[marcovarrone]

Hi @jmintch , thank you very much.

Your intuition is correct, it's the enrichment of cluster 4 around cluster 0, so the rows are the considered niche while the columns are the neighboring niches.
In fact, almost all the cells of the orange cluster are surrounded by purple cells (and other orange cells) -> high enrichment.
While the purple cells are both surrounded by orange cells and green cells (and other purple cells) -> positive enrichment but lower.

I hope it helped to clarify!

[jmintch]

perfect, thanks a lot for clarifying!