attempt to apply non-function while trying DEseq vignette sample data

[LanaG]

Hi,
I installed the DEseq version 1.0.6 from the repository to my mac, and am trying to use the example provided in the vignettes:

http://www.bioconductor.org/packages/2.7/bioc/vignettes/DESeq/inst/doc/DESeq.R

library("DESeq")
countsTable <- read.delim(system.file("extra/TagSeqExample.tab", package="DESeq"), header=TRUE, stringsAsFactors=TRUE, row.names="gene")
conds <- factor(c("T", "T", "T", "Tb", "N", "N"))
cds <- newCountDataSet( countsTable, conds )
cds <- estimateSizeFactors( cds )
cds <- estimateVarianceFunctions( cds )
res <- nbinomTest( cds, "T", "N")

every command seemed to work until it hits nbinomTest(), where I get an error: attempt to apply non-function. However when I type help(nbinomTest), the documentation of this function does pop up. Does anyone experience the same problem?

Also, when I checked what the variable cds is after applying estimateVarianceFunctions( cds ), I get the following output, is this normal???

Thank you so much!

> cds
CountDataSet (storageMode: environment)
assayData: 18760 features, 6 samples
element names: counts
protocolData: none
phenoData
sampleNames: T1a, T1b, ..., N2 (6 total)
varLabels and varMetadata description:
condition: NA
sizeFactor: NA
featureData: none
experimentData: use 'experimentData(object)'
Annotation:

[Simon Anders]

No clue what went wrong here, but maybe try again using the current release version of DESeq: http://www.bioconductor.org/packages...tml/DESeq.html

[LanaG]

Now I downloaded the newest DESeq 1.2.1. I have the same error again! But this time it occurs at estimateVarianceFunctions( cds ), even before nbionomTest(). What is wrong??

[RSK]

Ok, so I am having the same exact issue (on my mac), just with nbinomTest() and scvPlot(). I just downloaded the current version.

I would appreciate any insight.

Thanks!

[RSK]

The issue in my case seems to be only specific to the mac environment, when I run the sample on my Linux box, the issue goes away and everything works very smoothly.

[Wolfgang Huber]

Dear LanaG and RSK,

can you send the output of
sessionInfo()
One possibility is that the various R add-on packages on your computers are 'out of sync' and need to be all updated. (It's very unlikely a general issue of Mac vs Linux.)

If possible, please also always include a complete transcript of what you did (like LanaG in her first post), and the output of
traceback()

Best wishes
Wolfgang

[duygu]

any solution

Hi,
I am running into the same problem. Does anybody figure out how to resolve it ? Is it specific to Mac environment ? I appreciate any comments, ideas.

Thank you.
Duygu

This is how the code looks like:

counts <- read.table("All_regions.collapsed.txt", header = TRUE)
> conds <- c("e8","e9","e10","e10")
> cds <- newCountDataSet( counts[,1:4], as.factor(conds) )
> cds <- estimateSizeFactors( cds )
> cds <- estimateVarianceFunctions( cds )
Error: attempt to apply non-function

And this is how the traceback looks like:

traceback()
14: predict.locfit(fit, x[is.finite(x)])
13: predict(fit, x[is.finite(x)])
12: safepredict(fit, log(q))
11: pmax(safepredict(fit, log(q)) - xim * q, 1e-08 * q)
10: cds@rawVarFuncs[[cond]](q)
9: FUN("e10"[[1L]], ...)
8: lapply(replicated, function(cond) cds@rawVarFuncs[[cond]](q))
7: object@rawVarFuncs[[rvfName]](c(1.5, 3.5, 7.3))
6: validityMethod(object)
5: identical(x, TRUE)
4: anyStrings(validityMethod(object))
3: validObject(cds)
2: `rawVarFuncTable<-`(`*tmp*`, value = c("e10", "_max", "_max"))
1: estimateVarianceFunctions(cds)

[Simon Anders]

Maybe something is strange with your count data. Could you please save you count table and your count dataset to a file (with

save( counts, cds, file="foo.rda" )

and send me the file by email? Then I can have a look.