ATAC-seq deconvolution

[seoyeonoh12]

Dear,

Is it possible to use EPIC for cell type deconvolution using ATAC-seq datasets?
It gives the following error when I provide bulk ATAC-seq matrix (rows=peaks, columns=sample/subject) as input.

Error in EPIC::EPIC(bulk = bulk.mtx, sigGenes = rownames(Signature)) :
There are only 0 signature genes matching common genes between bulk and reference profiles, but there should be more signature genes than reference cells

Thank you in advance.

Kind regards,
Seoyeon

[jracle85]

Dear Seoyeon,

Thank you for your message and interest in using EPIC with ATAC-seq data.

Currently, the reference profiles available in EPIC are based on RNA-seq data with gene names as row and so EPIC would not be able to match the peaks to genes names. This is why EPIC writes that there are 0 signature genes in common between the bulk and reference profile used.

However, it would be possible to use EPIC if you defined your own reference profiles with peaks. Maybe this is what you are trying to do as you redefine signature genes with the sigGenes option. Maybe you forgot to also define the option reference, maybe using something like reference = list(refProfiles=Signature).

Please note that @aurelieGabriel is working on adapting EPIC to include reference profiles for ATAC-seq and these should be available in the coming months. We will let you know when these are available.

Best regards,

Julien