Hello,

I want to use DESeq2 for DE analysis. I tried to perform a nested comparison. My data has two different factors: the mice are either 1. Uninfected and Infected or 2. Susceptible and Resistant strain. I would like to know how to set the design and contrasts to determine the cell-line specific effect, which consists of:

Resistant.Infected - Resistant.Uninfected vs Susceptible.Infected - Susceptible.Uninfected.

Just to clarify, I am interested in determining those genes which have different gene expression response to infection depending on the mouse strain.

I tried

colData = data.frame(rownames(dataSet$fst.cls), dataSet$sec.cls, dataSet$cls)

view(colData)

condition        type         condition_type
1    Infected   Resistant     Infected.Resistant
2    Infected   Resistant     Infected.Resistant
3    Infected   Resistant     Infected.Resistant
4  Uninfected   Resistant   Uninfected.Resistant
5  Uninfected   Resistant   Uninfected.Resistant
6  Uninfected   Resistant   Uninfected.Resistant
7    Infected Susceptible   Infected.Susceptible
8    Infected Susceptible   Infected.Susceptible
9    Infected Susceptible   Infected.Susceptible
10 Uninfected Susceptible Uninfected.Susceptible
11 Uninfected Susceptible Uninfected.Susceptible
12 Uninfected Susceptible Uninfected.Susceptible

    colnames(colData) = c("condition", "type", "condition_type")
    dds = DESeqDataSetFromMatrix(countData=dataSet$data.anot, colData = colData, design = ~condition_type)
    dds = DESeq(dds)

resultsNames(dds)

[1] "Intercept"                            "condition_typeInfected.Resistant"    
[3] "condition_typeInfected.Susceptible"   "condition_typeUninfected.Resistant"  
[5] "condition_typeUninfected.Susceptible"

    res = results(dds, contrasts = c(0,1,-1,-1,1))

But when I look into the res file, the only analysis it performed is

"condition_typeInfected.Resistant vs condition_typeUninfected.Susceptible"

I know that this way works but I want to know how to use a more flexible method for design and contrast:

dds = DESeqDataSetFromMatrix(countData=dataSet$data.anot, colData = colData, design = ~type+condition + condition:type)

Thank you in advance.

hi,

Have you looked over the DESeq2 vignette section on interactions? What you want to do is to have a design with ~condition + type + condition:type. See the vignette section for details on the interpretation. Also there are examples in ?results.