Question: DESEQ2 Design formula testing cell type across condition
gravatar for Barry Digby
21 months ago by
Barry Digby630
National University of Ireland, Galway
Barry Digby630 wrote:


I am attempting to compare the following samples in DESEQ2 that follow and experimental design of:

condition    celltype
Ctrl          A
Ctrl          B
Ctrl          C
Ctrl          D
CYTO          A
CYTO          B
CYTO          C
CYTO          D

(Replicates left out for simplicity sake)

The researcher has requested I compare cell types to each other (A vs. B, A vs. C, A vs. D, B vs. C etc). I know how to compare Cell type A vs Cell type B for Control and CYTO, separately. I did this by merging condition and celltype into a new column called group, and called

 results(dds, contrast=c("group", "ACtrl", "BCtrl"))

However I would like to compare A vs. B across condition. Reading online, I have found this might be what I need to test genes which respond differently to cell type across condition:

design = ~ condition + celltype + condition:celltype

>resultsNames:  # "C" is the reference
[1] "Intercept"                     "condition_CYTO_vs_Ctrl"        "celltype_A_vs_C"          
[4] "celltype_B_vs_C"        "celltype_D_vs_C"        "conditionCYTO.celltypeA"   
[7] "conditionCYTO.celltypeB" "conditionCYTO.celltypeD"

Assuming this is correct, my confusion lies in calling the correct comparison in results. Do i need to use "name", "contrast", "list", "contrast=list" ?

Thank you in advance

rna-seq design deseq2 • 895 views
ADD COMMENTlink modified 21 months ago by Biostar ♦♦ 20 • written 21 months ago by Barry Digby630

In the first scenario where you created a new column, you could use contrast but in the 2nd scenario you have to use name from resultsNames. To compare A vs B, I would suggest revel to make A reference.

ADD REPLYlink written 21 months ago by prabin.dm220

Yeah I think I had gone too far,

design = ~ condition + celltype + condition:celltype

Is not what I was after. Leaving out condition:celltype allowed me to control for the effects of condition. My phrasing was unclear -- instead of 'across condition', I should have said controlling for condition. Very sloppy post, my bad.

ADD REPLYlink written 20 months ago by Barry Digby630
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1636 users visited in the last hour