Question: DESeq2 for Cut&Run and differential enrichment testing
gravatar for MPerch
14 days ago by
MPerch0 wrote:

Hi there,

I wish to compare differential enrichment of a histone modification (H3K9me2) across 5 replicates of control and treated samples, for each of two cell types. This modification is very broad and not amenable to peak analysis. Therefore I have used featurecounts to count up reads falling in gene-bodies for each gene and sample, and now have a counts matrix of K9me2 enrichment at each gene.

I would like to use this counts matrix in DESeq2 for differential enrichment analysis, as if treating it as RNA-seq in essence. However one issue I have is that for Cut&Run and for analysis of global changes in enrichment, I first normalize the data to the number of mapped reads generated from S.cerevisiae spike-in, within each sample. This means my data table now contains spike-in normalized values and not raw counts, which I know DESeq2 normally requires for DEseq(). If I round my values so that no error is thrown, is it OK to continue to use DEseq() to find genes with differential enrichment of the modification (and what would you recommend to set for the betaPrior?) If not advisable, is there a better way to allow normalization to spike-in values plus differential testing with DESeq2?

Many thanks for any help.


ADD COMMENTlink modified 14 days ago by Michael Love20k • written 14 days ago by MPerch0
gravatar for alex.gos90
14 days ago by
alex.gos9010 wrote:


I was recently very happy to find RUVSeq ( for normalisation prior differential analysis. They have a method specifically developed for spike-in normalisation (RUVg), and provide further advice how to proceed with DEseq2.

Maybe this helps you. 



ADD COMMENTlink written 14 days ago by alex.gos9010

I should add that I have used it for ATACseq data, but I do not know how it performs with broad ChIPseq. 

ADD REPLYlink written 14 days ago by alex.gos9010
gravatar for Michael Love
14 days ago by
Michael Love20k
United States
Michael Love20k wrote:

Do not normalize the data for sequencing depth and provide to DESeq2. This point is made frequently in our documentation.

You can instead do:

dds <- estimateSizeFactors(dds, controlGenes=...)
dds <- DESeq(dds)


ADD COMMENTlink written 14 days ago by Michael Love20k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 235 users visited in the last hour