cell_signaling function "paracrine" vs "autocrine"

[aguerosilva]

Hi I am working on my single-cell RNA seq dataset that is pre-processed by using Seurat pipeline to analyze intercellular interactions. When I run my data by cell_signaling function, I realized that when I chose "autocrine" option as an input, it gives markedly more number of interactions, something from 30 to 300 increase of interactions, than "paracrine" option. I am aware that "autocrine" option incorporates the result of both autocine and paracrine signalings. However, since even interactions between different cell types (paracrine signaling) are also markedly increased, this doesn't explain such huge increase in interactions. It would be great if you could comment on what is the potential problem here. Here are some of my codes and output in case any reference is needed.

`# scRNA (pre-processed seurat object) cluster <- as.numeric(Idents(scRNA)) data <- data.frame(scRNA[['RNA]]@data) all.genes <- rownames(scRNA)

signal.paracrine = cell_signaling(data=data,genes=all.genes,cluster=cluster, int.type = "paracrine", species = "mus musculus") signal.autocrine = cell_signaling(data=data,genes=all.genes,cluster=cluster, int.type = "autocrine", species = "mus musculus") ` Output For "paracrine" option: No such file as table_dge_cluster 1.txt in the cluster-analysis folder No such file as table_dge_cluster 2.txt in the cluster-analysis folder No such file as table_dge_cluster 3.txt in the cluster-analysis folder No such file as table_dge_cluster 4.txt in the cluster-analysis folder No such file as table_dge_cluster 5.txt in the cluster-analysis folder No such file as table_dge_cluster 6.txt in the cluster-analysis folder No such file as table_dge_cluster 7.txt in the cluster-analysis folder No such file as table_dge_cluster 8.txt in the cluster-analysis folder Paracrine signaling: Checking for signaling between cell types No significant interaction found from cluster 1 to cluster 2 No significant interaction found from cluster 1 to cluster 3 No significant interaction found from cluster 1 to cluster 4 No significant interaction found from cluster 1 to cluster 5 No significant interaction found from cluster 1 to cluster 6 No significant interaction found from cluster 1 to cluster 7 No significant interaction found from cluster 1 to cluster 8 34 interactions from cluster 2 to cluster 1 10 interactions from cluster 2 to cluster 3 No significant interaction found from cluster 2 to cluster 4 1 interactions from cluster 2 to cluster 5 10 interactions from cluster 2 to cluster 6 5 interactions from cluster 2 to cluster 7 8 interactions from cluster 2 to cluster 8 5 interactions from cluster 3 to cluster 1 No significant interaction found from cluster 3 to cluster 2 No significant interaction found from cluster 3 to cluster 4 No significant interaction found from cluster 3 to cluster 5 No significant interaction found from cluster 3 to cluster 6 3 interactions from cluster 3 to cluster 7 No significant interaction found from cluster 3 to cluster 8 46 interactions from cluster 4 to cluster 1

For "autocrine" option: Autocrine signaling: Checking for cell/cell signaling: 366 interactions from cluster 1 to cluster 1 347 interactions from cluster 1 to cluster 2 387 interactions from cluster 1 to cluster 3 316 interactions from cluster 1 to cluster 4 303 interactions from cluster 1 to cluster 5 274 interactions from cluster 1 to cluster 6 340 interactions from cluster 1 to cluster 7 325 interactions from cluster 1 to cluster 8 344 interactions from cluster 2 to cluster 1 309 interactions from cluster 2 to cluster 2 329 interactions from cluster 2 to cluster 3 284 interactions from cluster 2 to cluster 4 286 interactions from cluster 2 to cluster 5 257 interactions from cluster 2 to cluster 6 296 interactions from cluster 2 to cluster 7 281 interactions from cluster 2 to cluster 8 326 interactions from cluster 3 to cluster 1 323 interactions from cluster 3 to cluster 2 342 interactions from cluster 3 to cluster 3 292 interactions from cluster 3 to cluster 4 277 interactions from cluster 3 to cluster 5 256 interactions from cluster 3 to cluster 6 299 interactions from cluster 3 to cluster 7 293 interactions from cluster 3 to cluster 8

Thank you!

[yingyonghui]

Hi @aguerosilva I just run into the same question. Have you solved the problem? Any suggestions would be appreciated!

Many thanks!

[SCA-IRCM]

Hi,

The answer you're looking for is in the UsersGuide.html:

The “paracrine” option looks for ligands expressed in cluster A and their associated receptors according to LRdb that are expressed in any other cluster but A. These interactions are labelled “paracrine”.

The “autocrine” option searches for ligands expressed in cell cluster A and their associated receptors also expressed in A. These interactions are labelled “autocrine”. Additionally, it searches for those associated receptors in the other cell clusters (not A) to cover the part of the signaling that is “autocrine” and “paracrine” simultaneously. These interactions are labelled “autocrine/paracrine”.

Therefore you will find about 10 times more interactions with the "autocrine" option.

Hope this helps.

Thanks for using SingleCellSignalR!

SCA

[yingyonghui]

@SCA-IRCM Thanks for your explanation.

However I 'm still confused about the markedly increased number of interactions with the “autocrine” option.

As far as I understand, the "autocrine/paracrine" interactions contained the "autocrine" interactions and the"paracrine" interactions, so the number of "autocrine/paracrine" interactions with the "autocrine" option should equal to the sum of number of "autocrine" interactions with the "autocrine" option and the number of "paracrine" interactions with the "paracrine" option.

Thanks.

[SCA-IRCM]

Hi,

Actually, "autocrine/paracrine" covers interactions that are both "autocrine" and "paracrine", it is when a cell type produces a ligand for which it have the associated receptor and another cell type also have this receptor.

"autocrine" is when a cell type produces a ligand and only this cell type have the associated receptor.

"paracrine" is when a cell type produces a ligand and it doesn't have the associated receptor but one or several other cell types have it.

This is why the number of "autocrine/paracrine" interactions with the "autocrine" option is not equal to the sum of number of "autocrine" interactions with the "autocrine" option and the number of "paracrine" interactions with the "paracrine" option as it describes different types of interactions.

I hope this is clear for you now.

Feel free to respond if it's not.

Thanks for using SingleCellSignalR!

SCA