RNA seq

Question

[s.haefliger]

Dear Michael

I am contacting you because I saw on BioStar that you have a very deep knowledge in the field of RNA-seq.

Note that I am a novice in bioinformatics. My situation: I have raw counts for the tumor I am interested in (however no non-tumor control). However, I would like to control the expression with non-tumour tissue (muscle tissue). To do this, I would like to take the raw counts from GTEx, merge them with the raw counts from my tumour, apply CombatSeq to minimise the batch effect, then use DESeq2 and GSEA to see the differences in expression. Does this make sense to you? Is this method correct? I have bulk RNA seq data.

Thanks in advance and best regards

Answer 1

[Friederike]

If you completely confound condition and technical batch effect (in your case, tumor = batch 1 and control = batch2), there's no way to distinguish the technical effect from the biological effect of interest. I.e. if you apply CombatSeq to remove the batch effect, you will also remove all biologically meaningful differences between tumor and control.