How to deal with samples partially without replicates in DESeq?

[zatoichi888]

I have four groups of samples, 2 groups with replicates and 2 groups without. The data frame is like this:
expt_design<-data.frame(row.name=colnames(colon)
+ condition=c("GF","SPF","9343","9343","SPT","SPT"))
Error: unexpected symbol in:
"expt_design<-data.frame(row.name=colnames(colon)
condition"
> expt_design<-data.frame(row.names=colnames(colon),
+ condition=c("GF","SPF","9343","9343","SPT","SPT"))
> expt_design
condition
GF GF
SPF SPF
X9343.1 9343
X9343.2 9343
SPT.1 SPT
SPT.2 SPT
> conditions=expt_design$condition
> conditions
[1] GF SPF 9343 9343 SPT SPT
Levels: 9343 GF SPF SPT

When I tried to run it gave me this error message:
>cds=estimateDispersions(cds)
Error in .local(object, ...) :
None of your conditions is replicated. Use method='blind' to estimate across conditions, or 'pooled-CR', if you have crossed factors.

Plz help me to identify the problem and tell me how to deal with partially replicated samples

[dpryan]

Did you read the error message, the fix is contained within it. Also, you should read the DESeq tutorial.

[Simon Anders]

You might have confused DESeq by passing your "condition" data as a data.frame rather than as a factor. Try again using

cds <- newCountDataSet( countTable, exptDesign$condition)

And consider switching to DESeq2. It not only has improved power; we have also improved the interface to avoid problems like this one.