RNAseq one control two conditions, shared and exclusive genes

Question

[matteo.levorato]

Hi,

I am struggleling with RNAseq. I have 1 control and two conditions and I want to determine shared genes between the two conditions (compared to control) and exclusive genes for each control. Here my question:

• which is the pipeline (or workflow) to follow to discover a. shared genes between the two conditions; b. exclusive genes for each conditions?

Thnx

Answer 1

[Mohamed Abderrahmane]

I think it would be pertinent to use DESEq2 to perform two comparisons: the first one between the control group and condition 1, and the second one between the control group and condition 2. Later, you can analyze the genes that are differentially expressed in these two comparisons by Rstudio. If I were in your position, that's the first approach I would consider. This suggestion comes from someone who is relatively new to transcriptomic analysis.

[matteo.levorato]

thank you for your comment, but yes running DESeq2 is the starting point.

   dds <- DESeqDataSetFromMatrix(countData=RawCounts,
                               colData=sampleTable,
                               design= ~ Condition)

in the matrix 'RawCounts' are present the counts for the control, condition1 and condition2. And here starts the doubts: should I run this two times, one per condition? like splitting the initial matrix in 2, one with control and condition1 and another with control and condition2?

However, after runed DESeq2 with one matrix with all the conditions, I did:

    res.shr.I <- lfcShrink(dds, coef = "Condition_Condition1_vs_Control",type="apeglm")
    res.shr.II <- lfcShrink(dds, coef = "Condition_Condition2_vs_Control",type="apeglm")

in this way I have DEGs for the two condition, but I still don't know which are shared and which one are exclusive...

[Mohamed Abderrahmane]

In your table RawCounts, keep only the columns corresponding to condition 1 and the control. Edit the table sampleTable to align with your RawCounts table. Then, execute the remaining lines to identify differentially expressed genes between the control and the condition 1. Repeat this process for the second condition and you will have the differentially expressed genes between the control and the condition 2.

[matteo.levorato]

but why not a single matrix with the 3 different condition and then by lfcShrink function get the DEGs of each condition?

[Mohamed Abderrahmane]

Yes, why not? If it works for you like that, bingo!

[matteo.levorato]

I mean, which is the difference between doing two separated analysis or doing a one single analysis since the results are slightly different.

[Mohamed Abderrahmane]

I'm not entirely certain about the difference, but I assume it yields similar results. The method you're currently employing seems faster and more directly automated by DESEq2.