I use DEseq2 for classical analysis (condition A vs condition B) but I need now to use more complex formula and I'm not sure of what to do.
I have 3 conditions (with 4 replicats), without treatment (NT), with treatment A and with treatment B. However, treatment comes with nano. I am interested to analysis effect of treatment A and B and want to « remove » effect of nano.
Here is the description of my samples :
samplename filename condition nano
104 104 NT nano
105 105 NT nano
106 106 NT nano
107 107 NT nano
108 108 A nano
109 109 A nano
110 110 A nano
111 111 A nano
112 112 B nano
113 113 B nano
114 114 B nano
115 115 B nano
100 100 NT no_nano
101 101 NT no_nano
102 102 NT no_nano
103 103 NT no_nano
I don't know what I need to use in my formula. interaction ? Contrast ?
Thanks in advance for any help,
S.
I want to select DE genes between NT-nano vs A-nano but don't keep DE genes due to nano.
A basic method to do that may be to compute DE genes between NT no-nano and NT nano, and then with gene list intersection remove those genes to the NT-nano vs A-nano list. But I think it's possible to do better.
I hope I'm clear on my wish.
Thanks.
The basic method is what I would recommend.
If you had A and B samples in "no nano", then you could use a simple interaction model to test if the A vs NT effect was larger or smaller in nano vs no nano (an interaction term is often used to test for differences in treatment effects across groups, as diagrammed in the DESeq2 vignette), but for your experimental design it's not possible to perform this test.
Thank you for your answer.
Unfortunatly, I haven't no nano data for A and B samples, because we need nano to drive treatment into the cell.
I think it was possible to "normalised" and filtered fold change of NT-nano vs A-nano DE genes with fold change of NT no-nano vs NT nano.
For example, if the gene X have a log2FC of 1 between NT no-nano vs NT nano and a log2FC of 3 between NT-nano vs A-nano, the "real" effect of treatment A is a log2FC of 2.