DEXSeq log2 fold change error

[agne]

Hello,
It is my first time to use the DEXSeq package. I have performed all the steps according to the vignette and so far everything was working fine. However, I get the following error and warning messages after estimatelog2FoldChanges command:

Error in `row.names<-.data.frame`(`*tmp*`, value = c("geneID", "exonID", :
duplicate 'row.names' are not allowed
In addition: Warning messages:
1: In is.na(x) : is.na() applied to non-(list or vector) of type 'NULL'
2: non-unique value when setting 'row.names': ‘log2fold(/)’

Needless to say, no log2fold values are calculated. The head of the DEUresultTable has exactly the same structure as presented in the vignette, therefore I have trouble understanding, which row names are out of the required format or why this error only appears at this step.

I would really appreciate the help!

[agne]

problem solved

I managed to solve the problem.

[Rhizosis]

Whats the trick ?

[agne]

The problem was that the values of the grouping parameter (the "condition" according to the vignette) have to be factors. If they are not, then just converting them to factors is enough. Therefore the error message is somewhat misleading for an R novice like me. Hopefully it is helpful to others

[tankman]

hi agne,

It would be really helpful to me if you could include exactly how you converted to factors to get estimatelog2FoldChanges to work.

Thanks!

tm

[dpryan]

See the R factor() command.

[tankman]

Yes, I know what R factor() does -- the issue is on what I apply this to. If I apply this to sample$condition from the vignette -- as implied above -- it doesn't work, so I was hoping to learn the exact command to use.

[dpryan]

Define "it doesn't work".

[tankman]

Initial problem:
-----------------
> ecs <- estimatelog2FoldChanges(ecs)


Error in `row.names<-.data.frame`(`*tmp*`, value = c("geneID", "exonID", :
duplicate 'row.names' are not allowed
In addition: There were 50 or more warnings (use warnings() to see the first 50)
--------------
Attempted resolution:

> ecs <- estimatelog2FoldChanges(ecs, denominator = as.factor(as.character(design(ecs, drop = FALSE))[1]) )
Error: any(colnames(alleffects) %in% denominator) is not TRUE
In addition: There were 50 or more warnings (use warnings() to see the first 50)

---------------------

Basic question: what do I turn into a factor and thus how exactly do I modify the ecs object so that log2foldchanges won't have these redundant labels?

[dpryan]

The first error is something you'll need to look into more. You really shouldn't have any duplicate row names.

For the second error, just reset colData(ecs) to contain factors. The error you're seeing is due to that.

[tankman]

There is one error: estimatelog2FoldChanges does not work.

It doesn't work since it thinks there are duplicate row names, which is not true:

> any( duplicated( paste( geneIDs(ecs), exonIDs(ecs), sep=":") ) )
[1] FALSE
> any( duplicated( featureNames(ecs)) )
[1] FALSE

Therefore we are back to the original problem, for which I thought turning something to factors as indicated in the thread above was a solution.

[dpryan]

It doesn't help that you're using an old version (estimatelog2FoldChanges has been deprecated).

[tankman]

Well, I wish I could switch at this stage but I can't. The irksome thing here is that the above people already solved it and it's some idiotic R-specific inanity that takes up so much time.

Do you think there's a solution along the lines of what they say above?

[dpryan]

That'll probably not solve the base problem. Which version of DEXSeq are you using (just post the output of sessionInfo())?

[tankman]

Can you guess what they "factorized" above to solve this problem? I can't.

> sessionInfo()
R version 3.0.3 (2014-03-06)
Platform: x86_64-unknown-linux-gnu (64-bit)

locale:
[1] C

attached base packages:
[1] parallel stats graphics grDevices utils datasets methods
[8] base

other attached packages:
[1] DEXSeq_1.8.0 Biobase_2.22.0 BiocGenerics_0.8.0

loaded via a namespace (and not attached):
[1] Biostrings_2.30.1 GenomicRanges_1.14.4 IRanges_1.20.7
[4] RCurl_1.95-4.1 Rsamtools_1.14.3 XML_3.98-1.1
[7] XVector_0.2.0 biomaRt_2.18.0 bitops_1.0-6
[10] hwriter_1.3 statmod_1.4.18 stats4_3.0.3
[13] stringr_0.6.2 tools_3.0.3 zlibbioc_1.8.0