Gene-Scoring Table for Cell Type Enrichment

[pmayog]

Hello,

I am trying to retrieve a list of genes that contribute to the enrichment of each cell type. I came across some related issues, and I noticed that @bschilder previously mentioned the potential implementation of a gene-scoring table.

I assume this feature has already been implemented since I obtain a gene-associated table containing various columns of information (see attached image). Could you clarify what each column represents?

Thank you for your help!

[Al-Murphy]

Hey! It would be helpful if you could share a minimal, reproducible example explaining what you are trying to achieve - this should include the code to run EWCE. Without this it is not easy to help.

However, by "retrieve a list of genes that contribute to the enrichment of each cell type" - you may mean specificity of gene expression of genes to each cell type which you can get from the reference CTD used:

#get reference ctd from EWCE - any can be used but this is the one in the vignettes
ctd <- ewceData::ctd()
#get gene set of interest - replace this with the list of genes you want to test
hits <- ewceData::example_genelist()
#run EWCE analysis
#.....
#check which genes in gene list are driving enrichment in a cell type
#We can do this by inspecting the cell types specificity:
#The [[1]] chooses the level from high-level cell types down to cell subtypes
spec_matrix <- ctd[[1]]$specificity
#filter to the gene list of interest
#so the gene list sample I'm using is human whereas the CTD is mouse so convert
#the gene list to mouse using ont to one orthologues to query - EWCE does this
#with orthogene under the hood
#NOTE this may not be a necessary step for you, it only is if your gene list
#species differs to the ctd species
mapping <- orthogene::convert_orthologs(hits,input_species="human",
                                         output_species="mouse")
#mappings are stored in rownames
#make sure to filter to those present in the specificity matrix
mapping_filt <- mapping[rownames(mapping) %in% rownames(spec_matrix),]
hits_spec_matrix <- spec_matrix[rownames(mapping_filt),]
#finally to make things easier, we can rename the genes with their human one to 
#one mapping gene names
rownames(hits_spec_matrix) <- mapping$input_gene
#now you can print to see the specificity for each cell type or filter to
#the cell type of interest
hits_spec_matrix

> hits_spec_matrix
        astrocytes_ependymal endothelial-mural interneurons  microglia oligodendrocytes pyramidal CA1 pyramidal SS
Apoe              0.38862482        0.02497230  0.004412664 0.54516091       0.01998925   0.010850743  0.005989320
Bin1              0.07446817        0.10582173  0.082824951 0.20635581       0.26679272   0.161361620  0.102374994
Clu               0.75378227        0.02913857  0.067687291 0.03678649       0.03044411   0.036988511  0.045172747
Abca7             0.14920779        0.23350524  0.134269608 0.11138210       0.07641927   0.120191672  0.175024296
Cr1l              0.07440098        0.48934507  0.027425609 0.19438851       0.11073961   0.050810478  0.052889741
Picalm            0.07905728        0.20511812  0.072638785 0.24216689       0.21910989   0.092230826  0.089678196

Have a go at running this example and if that doesn't answer your question provide a minimal, reproducible example explaining what you are trying to do.

Thanks,
Alan.