How should I assign a pathway/pathways to the target genes when running nichenetr? about nichenetr

[Eisuan]

Dear user,
in order to help you with your requests, I'll require additional information:

1. How did you generate your list of target genes? Are those putative target genes of ligands (based on NicheNet predictions)?
2. Is your data bulk-RNA seq or sc?

Unfortunately, we do not provide enrichment procedures based on target genes or predicted ligands in our package yet.

[pariaaliour]

Thanks for your reply,
My data is single nucleus dataset. I followed the sender-focused approach. And the code I used to generate the target genes was:

active_ligand_target_links_df = best_upstream_ligands %>% lapply(get_weighted_ligand_target_links, geneset = geneset_oi, ligand_target_matrix = ligand_target_matrix, n = 200) %>% bind_rows() %>% drop_na()
active_ligand_target_links = prepare_ligand_target_visualization(ligand_target_df = active_ligand_target_links_df, ligand_target_matrix = ligand_target_matrix, cutoff = 0.33)

Moreover, if there is no function for enrichment analysis do you recommend any way to do so out of nichenetr (as you had one case in tutorial that target genes are the genes of a specific pathway)?
I appreciate your help on this!
Paria

[Eisuan]

Dear Paria,
In this case, I suggest utilising an over-representation analysis approach (enrichment procedures based on Fisher's exact test / hypergeometric distribution). This kind of approach requires the definition of a gene universe, which is the set of all the possible genes that might have been chosen as "interesting ones" or "not interesting ones" (e.g. in differential expression analysis, the former genes are those "differentially expressed" and the latter ones are "not differentially expressed").
In your case, your universe set is composed of all the target genes described in the ligand-target matrix.

To perform the analysis, you can use TopGO since it conveniently allows you to adjust for the gene universe. This package also supports enrichment procedures that leverage limitations that could arise if you use hierarchical structured annotations, such as GO-terms (e.g. parent-child terms relationships and information redundancy. Please, refer to Alexa 2006 for more info). I can not help suggesting specific sources for pathway annotations, since their quality might be subjective.

Best regards,
Daniele