Question

How to get TPM / FPKM after batch correction with DESeq2?

0

Entering edit mode

vk • 0

@cfe7a460

Last seen 12 months ago

Europe

I'm trying to adjust batch effect using deseq2 limma::removeBatchEffect like below:

###### Batch Correction with limma removeBatchEffect #######

dds <- DESeqDataSetFromMatrix(countData = data,
                              colData = coldata,
                              design = ~ Samplebatch + cond)

dds <- DESeq(dds)
dds

dds$batch <- as.numeric(dds$Samplebatch)
dds$cond
dds$batch

## vst after adding batch information
vsd_abc <- vst(dds, blind = FALSE)
vsd_abc$Samplebatch
vsd_abc$batch

design0 <- model.matrix(~cond, colData(vsd_abc))
assay(vsd_abc) <- limma::removeBatchEffect(assay(vsd_abc), vsd_abc$batch, design=design0) ### Batch Correction

After batch effect correction I would like to get TPM / FPKM for performing ssGSEA and various other analyses which are normalized for the length of the gene.

Is it possible to get TPM / FPKM after batch effect removal using limma:removeBatchEffect?

Any help is appreciated. Thank you !!

DESeq2 BatchEffect tpm fpkm • 1.4k views

ADD COMMENT • link 12 months ago vk • 0

score 1 · Answer 1 · 2023-04-16

1

Entering edit mode

ATpoint ★ 4.0k

@atpoint-13662

Last seen 20 hours ago

Germany

Use DESeq2::fpkm(), convert to log2 and correct these values with limma. Or correct any other normalized counts on log2 scale as you like.

ADD COMMENT • link 12 months ago ATpoint ★ 4.0k

0

Entering edit mode

thank you for your reply. So, first I can convert uncorrected counts to FPKM and then log2 scale and use it for Batch correction with limma::removeBatchEffect? Can I use any log2 scaled expression instead of FPKM like CPM or TPM?

ADD REPLY • link 12 months ago vk • 0

1

Entering edit mode

Yes to all.

ADD REPLY • link 12 months ago ATpoint ★ 4.0k

0

Entering edit mode

Thank you very much !!

ADD REPLY • link 12 months ago vk • 0