Bulk RNA-seq in disease vs controls: deconfound differential gene expression by estimated cell type proportions?

1

Entering edit mode

2.2 years ago

Peter ▴ 10

I have bulk RNAseq data for patients with a disease, and for control patients.

Cell deconvolution (using a single-cell RNA-seq reference - a separate study) shows that cell type proportions are changing a lot in cases vs controls. In addition, differential gene expression analysis (DESeq2) highlights many genes which vary in cases vs controls.

Is there a way to 'deconfound' the differential gene expression measures by the known changes in the cell type proportions? One option I thought of is to include the proportion of each cell type as covariates in DESeq2 but I wasn't 100% sure if that was valid.

The ultimate aim is to be able to say whether gene expression changes in disease vs controls are happening in specific cell types. Do you think this is possible ?

Thank you very much!

expression gene deconvolution cell differential • 575 views

ADD COMMENT • link updated 13 months ago by biogrin ▴ 20 • written 2.2 years ago by Peter ▴ 10

0

Entering edit mode

Hey Peter, I was just thinking about doing the same in my analysis, but unsure whether it would work.
Did we end up going forward with this? Any advice you could share?

ADD REPLY • link 13 months ago by biogrin ▴ 20

Login before adding your answer.