edgeR pairwise comparison

Question

[rykerklie7]

I have a cell culture treated with two drugs A and B that was sequenced (DatasetA), and the same cell culture that is treated with just drug A (DatasetB).

DatasetA = Cell culture1 +Drug A +DrugB
DatasetB = Cell culture1 +DrugA

What would be the best way to do a differential expression between DatasetA and DatasetB to find the genes that are expressed due to the effect of DrugB alone.

What I did was to do a pairwise comparison between DatasetA and DatasetB, and do a venn to eliminate the genes that are in common between both datasets, and also those genes that are unique to DatasetB in order to get relevant ones to DrugB.

Is there a better way to approach this statistically?

Answer 1

[Mark]

Read the edgeR manual it has really good examples: https://www.bioconductor.org/packages/release/bioc/vignettes/edgeR/inst/doc/edgeRUsersGuide.pdf

I think your usage fits perfectly for chapter 3.2. Use the glm approach with glmQLFTest for best results.

Also I'm assuming you have 2+ biological replicates in each group.

[rykerklie7]

The examples that you point to does not correspond to this situation. This is an example of mixed cell lines. I know what glmQLFTest is used for.

[Mark]

I'm not understanding your question then. If you have two groups A and B, if you perform a comparison of A-B in edgeR then to this answers your question. What am I missing? You need to explain your experimental design in more detail.