Statistical analysisi in DESeq2 with different condition

Question

[michelafrancesconi9]

Hi, I did some analysis in DESeq2 to look at the different gene expression between different condition

condition_breakfast$group2 <-  factor(paste0(condition_breakfast$Braccio, condition_breakfast$Campione)
condition_breakfast$group3 <- factor(paste0(condition_breakfast$Braccio, condition_breakfast$Campione, condition_breakfast$Risposta))

dds <- DESeqDataSetFromMatrix(countData=raw, 
                                             colData=condition_breakfast, 
                                             design=~ group2)
dds <- DESe2(dds)
res_AL <- results(dds_tot, contrast=c("group2", "AL", "AL"))


dds_tot <- DESeqDataSetFromMatrix(countData=raw, 
                                             colData=condition_breakfast, 
                                             design=~ group3)

dds_tot <- DESeq(dds_tot)
res_totAL <- results(dds_tot, contrast=c("group3", "ALRD", "ALpCR"))

now I want to see if adding "Risposta" in the condition, so in the design, is significant or not to change the gene expression. How I can do it?

I look at the gene and they are different, in particulare in dss_tot they are more but I want to understand if is significant or not.

Thank you

Answer 1

[Shred]

What you need to do is a Likelihood Ratio Test, between a full model (including Risposta) and a reduced model (without Risposta).

Quoting the vignette:

DESeq2 offers two kinds of hypothesis tests: the Wald test, where we use the estimated standard error of a log2 fold change to test if it is equal to zero, and the likelihood ratio test (LRT). The LRT examines two models for the counts, a full model with a certain number of terms and a reduced model, in which some of the terms of the full model are removed.