DESeq2 analysis of a chromosome region

Question

[Aron]

I have a question about examining RNA-seq data for differential expression that I just haven't seen posted before and haven't seen similar in the literature. We have a hypothesis about disrupting chromatin in a certain region of a chromosome and disrupting local gene expression. Let's say we only care about the genes within a 500-kb region. Doing qRTPCR on every gene in the region in multiple tissues and replicates is just not feasible so we used global RNA-seq to measure gene expression. So, if we just want to know if local gene expression is altered by our genetic manipulation, is it appropriate to only include the genes in the region for DE analysis? We have no particular hypothesis about which genes will change expression, just that one or more will.

Answer 1

[BioinfGuru]

I would run the analysis pipeline as normal, with the full data set, then annotate the deseq2 results with genomic ranges, then subset the range I am interested in.

If I was to subset the data before:

• alignment: a magnitude of misaligned reads
• exploratory data analysis: it would be very difficult to to detect hidden sources of technical variation
• deseq2: I think this will affect normalisation by library size.