pathway for single cell RNA-seq

Question

[kanwarjag]

I have a single cell RNA-seq data form 10X genomics. I want to perform pathway analysis of top 50 genes of each cluster on the the basis of DE. These clusters represent cells with similar expression of genes. So in pathway analysis of these top 50 genes I want to identify what possible type of cells they may be. There are several types of pathway analysis- like IPA, Enrichr, David and so on. Any suggestion which tool will be the the best to at least suggest that clustered cells are of a particular cell type and has a specific pathways enriched in them. e.g if Stem cell pathway is enriched we may say that these are pluripotent cells. Any suggestion?

[tiago211287]

Did you get the fastq files? assuming that you do. From the single cell RNA-Seq Data to a pathway analysis is quite a path...way. A suggestion is to you look at https://pachterlab.github.io/kallisto/singlecell.html and http://satijalab.org/seurat/get_started.html

[kanwarjag]

I have already analyzed the data and has top differentially expressed genes specific for each cluster. Instead of using known markers I want to use enrichment/ pathway analysis that what type of cells are in each cluster. Apology If my original question was not explained clearly.

[GenoMax]

Please use ADD COMMENT/ADD REPLY when responding to existing posts to keep threads logically organized.

[sc_analysis]

Which tool did you use? I have the same query of identifying my cell types from my single cell data using pathway enrichment analysis.

[Madelaine Gogol]

Biologists around here seem to like IPA, but I haven't tried it myself.

Answer 1

[tiago211287]

There are many ways of doing enrichment analysis. I usually like the TopGO package because it let me construct my background.

Answer 2

[igor]

If you only take the top 50 genes, you might be missing a lot of information. Also, depending on what the other clusters are, the differentially expressed genes will change. You can take the average expression of all genes for each cluster and then use a cell type deconvolution tool. Check this previous discussion where there are a few great suggestions: Deconvolution Methods on RNA-Seq Data (Mixed cell types)