Error when run testForDEU DEXSeq

Question

[Lila M]

Hie everybody, I have this error when I've tried to run dsd = testForDEU(dsd)

Error in (function (classes, fdef, mtable)  : 
  unable to find an inherited method for function ‘sizeFactors’ for signature ‘"list"’

My complete code is:

meta <- read.delim(file.path("meta"), header=T)
rownames(meta) <- meta$sample_ID

countfiles <- list.files(pattern = "*.txt", full.names = TRUE)
names(countfiles) <- gsub(".txt", "", basename(countfiles))
flattenedFile = list.files(pattern="gff$", full.names=TRUE)
basename(flattenedFile)
dsd <- DEXSeqDataSetFromHTSeq(countfiles = countfiles, sampleData = meta, 
                              design = ~ sample + exon + levels:exon, flattenedfile = flattenedFile)

genesForSubset <- unique(geneIDs(dsd))
dsd <- dsd[geneIDs(dsd) %in% genesForSubset,]
colData(dsd)
dsd = estimateDispersions(dsd)
dsd = estimateSizeFactors( dsd )
dsd = plotDispEsts(dsd)
dsd = testForDEU(dsd)

The flattened file was created using

 python dexseq_prepare_annotation.py gencode.v27.annotation.gtf annot.hg38.DEXSeq.gff

and the counts:

 ls *.bam.sam | xargs -I {} -n 1 python dexseq_count.py -p yes -s reverse -r pos annot.hg38.DEXSeq.gff {} {}.txt

Any suggestion or idea to solve my problem?

Thanks!

[andrew.j.skelton73]

Check str(dsd) after calling DEXSeqDataSetFromHTSeq, and again when subsetting. I suspect you've somehow made that object into a list rather than a DESeq2 object.

[michael.ante]

I think you forgot to call dsd = estimateSizeFactors( dsd )

[Lila M]

I forgot to include it in the code, but I've called it as well, thanks!

[michael.ante]

Did you call it in that order? Or to be more precise: after which call did the error message apear?

[Lila M]

Yes, I called it in that order, and the error appear after dsd = testForDEU(dsd)

[michael.ante]

AFAIK, you need to estimate the SizeFactors before estimating the dispersion.