assign conditions in DESeq2

Question

[yjliu1127]

Hi everyone,

I have one question about the command line for assigning conditions in DESeq2. The following are the count table I got. I want to assign WT6A, WT6B and WT6C as controls (3 replicates) and KO6A, KO6B and KO6C (3 replicates) as treated samples. column 1 is the gene name. column2-4 will be controls and column5-7 will be the treated samples.

• WT6A WT6B WT6C KO6A KO6B KO6C
• dnaA 6594 7412 3401 9067 7344 3803
• dnaN 4274 5048 3680 7368 7300 6571

How should I write the command line for this?

Thanks, Yong

[Devon Ryan]

Do you already have everything in a single file or data.frame? If so, see help(DESeqDataSetFromMatrix).

[yjliu1127]

Thanks Ryan! I'm really bad at coding. I checked the website. I still cannot figure out. My experiment is pretty straight forward. It's two-group comparison between control and treatment. Each group has 3 replicates. This is the code I typed in R:

coldata=data.frame(row.names=c('WT6A', 'WT6B', 'WT6C','KO6A','KO6B','KO6C'),
                   group=rep(c("WT6","KO6"),3,each=3),
                   treatment=rep(c("control","treated"),each=3)).

However, there is an error: Error in data.frame(row.names = c("WT6A", "WT6B", "WT6C", "KO6A", "KO6B", : row names supplied are of the wrong length.

Could you please let me know how to fix this problem?

[Devon Ryan]

coldata=data.frame(row.names=c('WT6A', 'WT6B', 'WT6C','KO6A','KO6B','KO6C'),
               treatment=rep(c("control","treated"),each=3))

The "group" part is redundant and was producing a vector of length 18 (rather than 6) anyway, due to having an extra 3 in there.

[yjliu1127]

Thanks! I have another mistake in the code.

> library("DESeq2")
> countMatrix <- read.table ("file.txt", header=TRUE, row.names=1)
> coldata=data.frame(row.names=c('WT6A', 'WT6B', 'WT6C','KO6A','KO6B','KO6C'),treatment=rep(c("control","treated"),each=3))
> coldata
     treatment
WT6A   control
WT6B   control
WT6C   control
KO6A   treated
KO6B   treated
KO6C   treated
> dds=DESeqDataSetFromMatrix(countData=countMatrix,colData=coldata,design=~treatment)
**Error** in DESeqDataSet(se, design = design, ignoreRank) : 
  some values in assay are not integers.

How can I fix this? and How can I set the control group as the basal level if I want to compare treated samples with control samples? Is the following code right?

> coldata$treatment<-relevel(coldata$treatment,ref='control')

Thanks a lot!

[Devon Ryan]

The control group is already the base level, "control" is lexicographically before "treated".

The error is due to there being non-integer values in countMatrix. Figure out what that is and fix it.

[yjliu1127]

Thank you very much! Yong