How to deconvolute whole blood and analyzed cell type separately for EWAS

Question

[simplitia]

Hi I have whole blood CpG from the 850K EPIC plateform. My goal is to try understand how deconvolusion works; that is would it be possible separate out the cpgs into different subtype and run a EWAS for each cell type. separately?

For example lets say I'm comparing group A to control using whole blood. I then run separate EWAS analysis on two of the major cell type, lets say T-Cell and PMBC. Is this possible? or is it more like I run an anlysis first to predict proportion and while running EWAS use that proportion as a covariate? Sorry this is new to me and I'm trying to understand it better.

Moreover, it would be great if someone can recommend a nice R package to do this would be great.

Thank you.

Answer 1

[Kevin Blighe]

There is a package in R to do this, called MethylDeconBloodSubtypes. I have not used this before but I have performed cellular deconvolution on gene expression data.

In a nutshell: a deconvolution method comes with a set of pre-defined signatures (of gene expression or methylation) for each cell-type. These signature are produced by analysing expression or methylation in known cell-types and at known concentrations. The method then checks the level of expression/methylation of these signatures in your own data and can, through this proces, deduce both the type of cells present and their concentrations. There are multiple ways of performing this.

You may read this work in order to better understand how to bring the EWAS part into this: Estimation of Cell-Type Composition Including T and B Cell Subtypes for Whole Blood Methylation Microarray Data

From what I understand, the deconvolution method will produce some form of score for each probe, will can hten technically be used forrr EWAS. I'm not sure of the specifics of the MethylDeconBloodSubtypes package though.

Kevin