Hi, I am running a DESeq2 analysis which includes an interaction term between treated_cells and contamination (contamination = 1- purity ). The DESeq design looks like this :

dds <- DESeqDataSetFromMatrix( countData = counts, colData = manifest, design = ~ contamination + treated_cells + contamination : treated_cells )

With contamination and treated_cells being two continuous variables.

My question is: being interested in the interaction between contamination and treated_cells, is there a way to discriminate down or upregulated genes that likely come from the contamination (or treated_cells) ?

Many thanks, Selma.

I don't really know what you mean by "come from the contamination". You can look at the main effects and the interaction, but I don't know if you can make other causal claims beyond seeing the the effect is modified when treatment and contamination are both present.