Hi,

My data set is similar to the described "two conditions, two genotypes with an interaction term" situation. I think I do follow the example (2) correctly. However, when extracting results for the condition (e.g. treatment) effect for genotype II (e.g. X-KD), resultsNames function does not show the expected output. I know there are similar posts around but I could not figure it out myself.

Here is my code:

> library(DESeq2)
> coldata
                transfection      treatment ID
Ctr_untreated_1-1      siCtr      untreated  A
Ctr_untreated_1-2      siCtr      untreated  A
Ctr_untreated_2-1      siCtr      untreated  B
Ctr_untreated_2-2      siCtr      untreated  B
Ctr_untreated_3-1      siCtr      untreated  C
Ctr_untreated_3-2      siCtr      untreated  C
Ctr_untreated_4-1      siCtr      untreated  D
Ctr_untreated_4-2      siCtr      untreated  D
X-KD_untreated_1-1       siX      untreated  E
X-KD_untreated_1-2       siX      untreated  E
X-KD_untreated_2-1       siX      untreated  F
X-KD_untreated_2-2       siX      untreated  F
X-KD_untreated_3-1       siX      untreated  G
X-KD_untreated_3-2       siX      untreated  G
X-KD_untreated_4-1       siX      untreated  H
X-KD_untreated_4-2       siX      untreated  H
Ctr_treated_1-1        siCtr        hormone  I
Ctr_treated_1-2        siCtr        hormone  I
Ctr_treated_2-1        siCtr        hormone  J
Ctr_treated_2-2        siCtr        hormone  J
Ctr_treated_3-1        siCtr        hormone  K
Ctr_treated_3-2        siCtr        hormone  K
Ctr_treated_4-1        siCtr        hormone  L
Ctr_treated_4-2        siCtr        hormone  L
X-KD_treated_1-1         siX        hormone  N
X-KD_treated_1-2         siX        hormone  N
X-KD_treated_2-1         siX        hormone  O
X-KD_treated_2-2         siX        hormone  O
X-KD_treated_3-1         siX        hormone  P
X-KD_treated_3-2         siX        hormone  P
X-KD_treated_4-1         siX        hormone  Q
X-KD_treated_4-2         siX        hormone  Q

> raw_matrix <- read.table("Sequencing/HTSeq-count_allgenes_mod_order.txt", row.names=1, header=T)
> colnames(raw_matrix)<-NULL
> dds = DESeqDataSetFromMatrix(countData = raw_matrix, colData = coldata, design = ~ transfection + treatment + transfection:treatment)
> ddsColl <- collapseReplicates(dds, dds$ID)
> dds.test = DESeq(ddsColl)
> resultsNames(dds.test)
[1] "Intercept"                             "transfection_siX_vs_siCtr"        
[3] "treatment_untreated_vs_hormone" "transfectionsiX.treatmentuntreated"

I was expecting to get the output "transfectionsiX.treatmenthormone" instead of "transfectionsiX.treatmentuntreated".

. Where do I miss something?

Thanks,

Zeynep

See the DESeq2 vignette section on setting factor levels.

Hi,

Thank you for the feedback. As I understood correctly, I can use relevel function in order to set the reference.

I did try this indeed before running the DESeq function (after collapsing replicates):

> ddsColl$transfection <- relevel (ddsColl$transfection, "siCtr")
> ddsColl$treatment <- relevel (ddsColl$treatment, "untreated")
> dds.test = DESeq(ddsColl)
> resultsNames(dds.test)
[1] "Intercept"                                  "transfection_siPLZF_vs_siCtr"             
[3] "treatment_hormonetreated_vs_untreated"      "transfectionsiPLZF.treatmenthormonetreated"

Here, I got the expected output from resultsNames.

Then in order to extract results for the treatment effect for X-KD, I wrote a similar code to the example (2):

> res <- results(dds.test, contrast=list(c("treatment_hormone_vs_untreated", "transfectionsiX.treatmenthormone"), alpha=0.05)

It gives this error:

Error in checkContrast(contrast, resNames) :
  'contrast', as a list of length 2, should have character vectors as elements,
see the manual page of ?results for more information

I don`t know where I miss something. Did I use relevel function wrong?

Thanks,

Zeynep

Thank you very much! It did solve the problem.