Remove batch effect for many RNA-seq

Question

[Shaway]

Hello every,

I a new comer for RNA-seq, here I have some questions for RNA-seq analysis.

I want to analyze about 10 type immune cells (more than 100 samples) and find the differential expressed genes. Now, I know that these data are from 12 different paper, I treat this as 12 batch then want to remove batch effect cause by different experiment.

Is it right for my understanding about batch effect?

Or I can do DE analysis just using TPM generated from raw counts.

Thank you, everyone!!

Answer 1

[Ram]

You should model the batch as a covariate in differential expression analysis. Avoid ComBat if you can, I don't think "removing" batch effect is all that effective. IMO.

You could also use ComBat on TPM, but be careful in interpreting the results.

[Shaway]

Thanks for your help!