[BioC] DEXSeq error during fold change calculation

[Alejandro Reyes]

Hi Beth,

Thanks for your email.  This error comes from the glm function, sometimes it fails due to really high dispersions. e.g. cases where you have 0 counts in an exon in almost all your samples and one sample with 1000 counts. In my experience this rarely happens so it should not be a problem for the analysis. Could you try updating to the newest version of DEXSeq? This version should convert these error into warnings, if this does not work, could you send me your ExonCountSet object to give a closer look to the error message?

Best wishes,
Alejandro Reyes


Date: Mon,  9 Apr 2012 10:05:52 -0700 (PDT)
From: "Beth [guest]"<guest at bioconductor.org>
To:bioconductor at r-project.org,elizzybethy at hotmail.com
Subject: [BioC] DEXSeq error during fold change calculation
Message-ID:<20120409170552.035491313D0 at mamba.fhcrc.org>


I am interested in detecting splicing differences between 2 conditions, with 5 biological replicates of each, and 	DEXSeq seems like a good tool for the job.


> ABsamples
     condition replicate
A013        A         1
A014        A         2
A015        A         3
A016        A         4
A017        A         5
B013        B         1
B014        B         2
B015        B         3
B016        B         4
B017        B         5

I generated an exon count matrix for non-overlapping Ensemblv64 exon parts using the provided dexseq_prepare_annotation.py and BEDtools. (I used Tophat for alignment.)

As far as I can tell, the ExonCountSet was created successfully from this input

> ABMatrix<- read.table(file="ABExonsMatrix.txt")
> ABgeneExonIDs<- read.table(file="ABExonIDs.txt")
> ABsamples<- read.table("ABsamples.txt")
> ABExon<- newExonCountSet(countData = ABMatrix, design = ABsamples, geneIDs = ABgeneExonIDs$V2, exonIDs = ABgeneExonIDs$V1)
I proceeded to this step:

> AB013to017<- makeCompleteDEUAnalysis(ABExon, nCores=22)
and the "Estimating Cox-Reid exon dispersion estimates" and the "Testing for differential exon usage" steps also looked like they worked (though there were a few errors like this for some of the exons "Error in chol.default(XVX + lambda * I) : the leading minor of order 10 is not positive definite", but I am pretty sure that isn't my underlying problem)

But then this happened:
Calculating fold changes using 22 cores. (Progress report: one dot per 100 genes)
Error in function (classes, fdef, mtable)  :
   unable to find an inherited method for function "fData", for signature "try-error"
In addition: Warning message:
In mcLapply(allecs, FUN = funtoapply, mc.cores = mc.cores) :
   all scheduled cores encountered errors in user code

I am not a very experienced R user, so I am not sure if this error stems from problems with my dataset, problems with the underlying R environment, or a problem with DEXSeq itself.

I would appreciate any thoughts and feedback you may have.  Please let me know if I should provide additional information and I would be happy to do so.

Thanks for your time,
Beth

  -- output of sessionInfo():


> sessionInfo()
R version 2.14.2 (2012-02-29)
Platform: x86_64-unknown-linux-gnu (64-bit)

locale:
  [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C
  [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8
  [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8
  [7] LC_PAPER=C                 LC_NAME=C
  [9] LC_ADDRESS=C               LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C

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

other attached packages:
[1] multicore_0.1-7     DEXSeq_1.0.2        Biobase_2.14.0
[4] BiocInstaller_1.2.1

loaded via a namespace (and not attached):
[1] hwriter_1.3    plyr_1.7.1     statmod_1.4.14 stringr_0.6    tools_2.14.2