Question

DESeq2 correction for biological replicate variance using design

0

Entering edit mode

lukas.krauss • 0

@6d5fa858

Last seen 16 months ago

Germany

Hello,

I have an issue with my RNAseq regarding the variance between my biological replicates.

If i include them as a factor of variance in my design, i can fix the issue nicely and looking at the PCA plot when removing"batcheffect" with:

mat <- assay(vsd)
mat <- limma::removeBatchEffect(mat, vsd$Biological_rep)
assay(vsd) <- mat

the samples cluster very nicely according to their different treatment conditions. enter image description here

My question now is if it is acceptable to do this or if this takes away the whole point of using replicates in the first place.

Thanks alot for your help and insight!

Lukas

DESeq2 • 701 views

ADD COMMENT • link 18 months ago lukas.krauss • 0

score 0 · Answer 1 · 2022-11-10

0

Entering edit mode

swbarnes2 ★ 1.3k

@swbarnes2-14086

Last seen 19 hours ago

San Diego

The typical way to deal with large donor effect like that in DESeq2 is not to alter the counts, but to include Biological_rep in the model. You won't be able to get a visual conformation it is working like you get with the PCA, though.

The other acceptable method is to use software called ComBat-seq. This will alter the counts, but in a way that is still compatible with DESeq.

ADD COMMENT • link 18 months ago swbarnes2 ★ 1.3k

0

Entering edit mode

Thank you very much for your quick reply.

By including it into my model you would mean adding it to the design like this?

ddsMat <- DESeqDataSetFromMatrix(countData = countmat,
                                 colData = df,
                                 design = ~Treatment+Biological_rep)

ADD REPLY • link 18 months ago lukas.krauss • 0