[Bioc-devel] baySeq countData class bug?

Tom Hardcastle tjh48 at cam.ac.uk
Mon Feb 23 16:48:18 CET 2015


Hi,

 From v2, baySeq has moved from storing data in matrices to arrays, to 
allow for multi-dimensional analyses. This looks like an unintended 
consequence of that change - I'll look into it and see if it's fixable, 
or whether it's indices only from now on.

Thanks for bringing it to my attention.

Cheers,

Tom

On 20/02/15 11:00, bioc-devel-request at r-project.org wrote:
> Message: 5
> Date: Fri, 20 Feb 2015 11:55:21 +0200
> From: Panagiotis Moulos<moulos at fleming.gr>
> To:"bioc-devel at r-project.org"  <bioc-devel at r-project.org>
> Subject: [Bioc-devel] baySeq countData class bug?
> Message-ID:<54E70489.8050108 at fleming.gr>
> Content-Type: text/plain; charset="UTF-8"
>
> Hello,
>
> In the previous Bioconductor release (2.14), a countData object from the
> baySeq package could be accessed by sample names. Now it seems that it
> can be accessed only by indices (version 2.0.50). Was this done on
> purpose or is it a bug unnoticed so far? I discovered this as baySeq
> fails with my package metaseqR. Some steps to reproduce:
>
> library(metaseqR)
> data("mm9.gene.data",package="metaseqR")
> result <- metaseqr(
>          counts=mm9.gene.counts,
>          sample.list=sample.list.mm9,
>          contrast=c("e14.5_vs_adult_8_weeks"),
>          libsize.list=libsize.list.mm9,
>          annotation="embedded",
>          id.col=4,
>          gc.col=5,
>          name.col=7,
>          bt.col=8,
>          org="mm9",
>          count.type="gene",
>          normalization="edger",
>          statistics="bayseq",
>          qc.plots="mds",
>          pcut=0.05,
>          fig.format="png",
>          export.what=c("annotation","p.value","fold.change"),
>          export.scale="log2",
>          export.values="normalized",
>          export.stats="mean",
>          export.where="~/bayseq_test",
>          out.list=TRUE
> )
>
> crashes with: Error in r[i1] - r[-length(r):-(length(r) - lag + 1L)] :
> non-numeric argument to binary operator
>
> and the problem stems from the countData class:
>
> object <- metaseqR::normalize.edger(mm9.gene.counts[,9:12],sample.list.mm9)
> CD <- new("countData",data=object,replicates=c(
> "e14.5","e14.5","adult_8_weeks","adult_8_weeks"))
> baySeq::libsizes(CD) <- baySeq::getLibsizes(CD)
>
> and then
>
> cd <- CD[,c("e14.5_1","e14.5_2")]
>
> crashes with
>
> Error in r[i1] - r[-length(r):-(length(r) - lag + 1L)] :
>     non-numeric argument to binary operator
>
> while
>
> cd <- CD[,1:2]
>
> works. In the version released with Bioconductor 2.14, this was not
> happening and the countData object could be accessed by sample names.
> Thanks in advance for informing me whether this is a bug or if it will
> remain like this from now on so I can update my code.
>
> Panagiotis
>
> -- *Panagiotis Moulos* Post doctoral researcher Biomedical Sciences 
> Research Center 'Alexander Fleming' Fleming 34, 16672, Vari, Greece 
> e-mail: moulos at fleming.gr <mailto:moulos at fleming.gr> [[alternative 
> HTML version deleted]]


-- 
Dr. Thomas J. Hardcastle
Senior Research Associate

Department of Plant Sciences
University of Cambridge
Downing Street
Cambridge, CB2 3EA
United Kingdom


	[[alternative HTML version deleted]]



More information about the Bioc-devel mailing list