Hi all-

I would like to be able to contrast the condition within a cell type (e.g. R1Vehicle vs. R1 Treated), but when I run DESeq with my group levels, it appears that the results from DESeq2 are already contrasting levels against a single level (P1Treated). Am I inputting the groupings wrong for my type and condition? Thanks in advance.

a <- read.csv("~/Desktop/BIX/SampleSheetBIX.csv", header=T)

   sampleName fileName type condition
1  42929veh_1   R1.txt   R1   Vehicle
2  42929veh_2   R2.txt   R1   Vehicle
3   42929tx_1   R3.txt   R1   Treated
4   42929tx_2   R4.txt   R1   Treated
5  20342veh_1   R5.txt   R2   Vehicle
6  20342veh_2   R6.txt   R2   Vehicle
7   20342tx_1   R7.txt   R2   Treated
8   20342tx_2   R8.txt   R2   Treated
9  40977veh_1   R9.txt   R3   Vehicle
10 40977veh_2  R10.txt   R3   Vehicle
11  40977tx_1  R11.txt   R3   Treated
12  40977tx_2  R12.txt   R3   Treated
13 54074veh_1  R13.txt   P1   Vehicle
14 54074veh_2  R14.txt   P1   Vehicle
15  54074tx_1  R15.txt   P1   Treated
16  54074tx_2  R16.txt   P1   Treated
17 99142veh_1  R17.txt   P2   Vehicle
18 99142veh_2  R18.txt   P2   Vehicle
19  99142tx_1  R19.txt   P2   Treated
20  99142tx_2  R20.txt   P2   Treated

dds <- DESeqDataSetFromHTSeqCount(sampleTable=a, design= ~type+condition)

dds$group <- factor(paste0(dds$type, dds$condition))

design(dds) <- ~ group

> dds$group
[1] R1Vehicle R1Vehicle R1Treated R1Treated R2Vehicle R2Vehicle R2Treated R2Treated R3Vehicle R3Vehicle R3Treated R3Treated P1Vehicle P1Vehicle P1Treated P1Treated
[17] P2Vehicle P2Vehicle P2Treated P2Treated
Levels: P1Treated P1Vehicle P2Treated P2Vehicle R1Treated R1Vehicle R2Treated R2Vehicle R3Treated R3Vehicle

dds2 <- DESeq(dds)

> resultsNames(dds2)
[1] "Intercept"                    "group_P1Vehicle_vs_P1Treated" "group_P2Treated_vs_P1Treated" "group_P2Vehicle_vs_P1Treated" "group_R1Treated_vs_P1Treated"
[6] "group_R1Vehicle_vs_P1Treated" "group_R2Treated_vs_P1Treated" "group_R2Vehicle_vs_P1Treated" "group_R3Treated_vs_P1Treated" "group_R3Vehicle_vs_P1Treated"

You can compare any two levels of group using the 'contrast' argument:

res <- results(dds, contrast=c("group", "...", "..."))

resultsNames(dds) just prints the coefficients that were fit, but you can use 'contrast' to make comparisons of interest.