DESeq Estimate Dispersion

[mattia]

Hi,
I'm starting to use DESeq package since few days, so I'm not very practice yet. I would ask two question.
First ten rows of my data file (TXT format) are presented below:

X6a X6b
ENSG00000000003 20 10
ENSG00000000005 0 0
ENSG00000000419 542 438
ENSG00000000457 468 338
ENSG00000000460 94 81
ENSG00000000938 19604 14802
ENSG00000000971 178 151
ENSG00000001036 559 335
ENSG00000001084 559 312

1st question: Is my data file in the right format?
2nd question:
I execute these commands:
>fuv6a6b<-read.table("C:/Programmi/R/R-devel/library/DESeq/FUV6/rawcount6a-6b.txt",header=TRUE,sep="\t",row.names=1)

>conds <- factor( c( "X6a", "X6b"))

>cds <- newCountDataSet( fuv6a6b, conds )

>cds <- estimateSizeFactors( cds )

>sizeFactors( cds )

>head( counts( cds, normalized=TRUE ) )

So far, so good! But when I execute this command (for estimate Variance):

>cds <- estimateDispersions( cds )

the program print this error message:
Error in rowSums(sapply(tapply((1:ncol(counts))[replicated_sample], factor(conditions[replicated_sample]), :
'x' must be an array of at least two dimensions

How can I solve this problem?

Thanks a lot.
Mattia

[Simon Anders]

Dear Mattia

your problem is that you do not have replicates. DESeq should have stated this in the error message; I have to check why the rather cryptic message appeared instead.

For you to go ahead, please read the section in the vignette about how to proceed without replicates and also look through the various threads here in SeqAnswers that discuss why your chances of getting any useful results from an RNA-Seq experiment without replication are rather slim.

Simon