Using Deseq2 for RNA-seq analysis of 4 different groups of cells

Question

[yeliezer21]

Hello, my name is Jonathan I'm new to R and to RNA-seq. I have a question regarding a project I'm working on. I want to compare differential expression of RNA within 4 different groups of cells (different tumors) each has about 35 people. I need to look for deferential expression of about 70,000 RNA. can I use DESEQ2 for that even though the cell are not paired or is there any other tool in order to do so? if someone can link me a manual for that kind of analysis I would be grateful. thank you for your support, Jonathan.

[ATpoint]

Can you add some details? So you have four tumor types. When you say "within" do you want DEGs separating subgroups within each entity, or do you want to compare different types of cancer? What does "70.000 RNA" mean, so do you need differential genes or transcripts?

Edit: Hope you get my point, if not please also comment. Terminology can sometimes be difficult if getting started with a new topic, there are no dumb questions, so please feel free to ask.

[yeliezer21]

yeah, I want to compare those different types of cancer (those are actually subgroups of he same cancer). I have 70,000 different genes and I want the differential expression between those 4 groups for those 70,000 genes (not within the same group but comparing the groups to each other). thank you for the help and sorry for the vague answer I'm new to this.

[ATpoint]

Are all these cancers from the same study, so is the wetlab processing the same? You can only meaningfully compare samples from the same lab. Can you confirm the data are not from different sources/labs/studies? In principle DESeq2 can do that, so comparing groups of samples, that is in fact the main use case, if I get you correctly.

[yeliezer21]

yeah, all of those samples are from the same research and the wetlab processing was the same. do you know of any tutorial that shows how to do that corectly?

Answer 1

[Carlo Yague]

If I understand correctly, you have four different tumors, but no reference healthy sample to compare to. In that case, you can take two different approaches with DESeq2 (given that they are comparable/from the same experimental batch, as ATpoint pointed out):

• Either make pairwise comparison of the different groups then combine the results (i.e, genes differentially expressed in tumor A vs B & A vs C & A vs D).

• Another possibility is to compare tumor A vs the average of B,C and D. This can be specified with:

  results(dds, contrast=list("tumorA", c("tumorB","tumorC","tumorD")), listValues=c(1, -1/3))

[yeliezer21]

exactly, so my options are to do this separately for each comparison (a\b,a\c,a d,b\c,b\d,c\d) or to do this with the average? do you know any tutorial that shows you how to do that? for example after I've compared each pair how can i combine the results? sorry if my questions are basic im willing to read on that subject if there's something you think i should.

[Carlo Yague]

I suggest that you start with this excellent DESeq2 tutorial. My suggestions above will fit after you reach the step with the results function (so after loading the data, defining the design, and running DESeq2).

after I've compared each pair how can i combine the results?

After you compare each pair, you will have a list of upregulated or downregulated genes for each pairwise comparison. Then you can just select genes that are upregulated or downregulated in all pairwise comparison.