Question

bug in "dispersion" argument in edgeR - rpkmByGroup

0

Entering edit mode

dario.beraldi • 0

@darioberaldi-6991

Last seen 2.1 years ago

United Kingdom

Hi- Trying to run the rpkmByGroup function from edgeR, I get the error below.

 rpkmByGroup(y, gene.length= rep(1000, nrow(y)))
Error in cpmByGroup.default(y = y, group = group, dispersion = dispersion,  : 
  formal argument "dispersion" matched by multiple actual arguments

I think it is because rpkmByGroup.default runs cpmByGroup with the dispersion argument specified twice:

rpkmByGroup.default
function (y, group = NULL, gene.length, dispersion = 0.05, offset = NULL, 
    weights = NULL, log = FALSE, prior.count = 2, ...) 
{
    if (is.null(group)) {
        group <- factor(rep_len(1, ncol(y)))
        levels(group) <- "AveRPKM"
    }
    z <- cpmByGroup(y = y, group = group, dispersion = dispersion, 
        dispersion = dispersion, offset = offset, weights = weights, 
        log = log, prior.count = prior.count, ...)
    if (log) {
        z - log2(gene.length/1000)
    }
    else {
        z/(gene.length/1000)
    }
}

Am I missing something or is this a bug? If the latter, I wonder if rpkmByGroup has been used at all and if it is recommendable to use.

Thanks!

sessionInfo()
R version 3.5.1 (2018-07-02)
Platform: x86_64-conda_cos6-linux-gnu (64-bit)
Running under: CentOS Linux 7 (Core)

Matrix products: default
BLAS: /export/projects/III-data/wcmp_bioinformatics/db291g/miniconda3/envs/20190619_brett_ko_PB0312700/lib/R/lib/libRblas.so
LAPACK: /export/projects/III-data/wcmp_bioinformatics/db291g/miniconda3/envs/20190619_brett_ko_PB0312700/lib/R/lib/libRlapack.so

locale:
 [1] LC_CTYPE=en_GB.UTF-8       LC_NUMERIC=C               LC_TIME=en_GB.UTF-8        LC_COLLATE=en_GB.UTF-8     LC_MONETARY=en_GB.UTF-8    LC_MESSAGES=en_GB.UTF-8    LC_PAPER=en_GB.UTF-8       LC_NAME=C                  LC_ADDRESS=C              
[10] LC_TELEPHONE=C             LC_MEASUREMENT=en_GB.UTF-8 LC_IDENTIFICATION=C       

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

other attached packages:
[1] ggplot2_3.1.0     edgeR_3.26.0      limma_3.40.0      data.table_1.12.0

loaded via a namespace (and not attached):
 [1] Rcpp_1.0.1       magrittr_1.5     splines_3.5.1    tidyselect_0.2.5 munsell_0.5.0    colorspace_1.4-1 lattice_0.20-38  R6_2.4.0         rlang_0.3.4      plyr_1.8.4       dplyr_0.8.1      tools_3.5.1      grid_3.5.1       gtable_0.3.0     withr_2.1.2     
[16] lazyeval_0.2.2   assertthat_0.2.1 tibble_2.1.3     crayon_1.3.4     purrr_0.3.2      glue_1.3.1       compiler_3.5.1   pillar_1.4.1     scales_1.0.0     locfit_1.5-9.1   pkgconfig_2.0.2

edger rpkm • 1.7k views

ADD COMMENT • link updated 5.5 years ago by rosehahn8583 • 0 • written 5.5 years ago by dario.beraldi • 0

score 1 · Answer 1 · 2019-06-20

1

Entering edit mode

Gordon Smyth 52k

@gordon-smyth

Last seen 1 hour ago

WEHI, Melbourne, Australia

Yes, that is a bug, which was introduced to edgeR in the last Bioconductor release. I have fixed it now in edgeR 3.26.5. Thanks for the heads-up.

It is curious that you are running versions of limma and edgeR from the Bioconductor 3.9 release under R 3.5.1. Bioconductor 3.9 is designed for R 3.6.0, and you won't be able to install the current versions of limma or edgeR unless you do run 3.6.0.

ADD COMMENT • link 5.5 years ago Gordon Smyth 52k

0

Entering edit mode

Hi Gordon- Thanks for quick reply, as always! About versions, I installed edgeR via conda/bioconda using conda install bioconductor-edger=3.26.0. It appears that the bioconda recipe for version 3.26.0 (here) requires limma >=3.40.0,<3.41.0.

ADD REPLY • link 5.5 years ago dario.beraldi • 0

0

Entering edit mode

I am able to reproduce this error using conda create --name edgeRenv bioconductor-edger.

It seems that Bioconda uses R from conda-forge (right now it uses R-3.5.1), and they are waiting for R-3.6.1 to be released before updating to that. This seems to be caused by the way Bioconda maintainers design their system.They do not add R-3.6.0 to their distribution, they wait for R-3.6.1.

Please try to install R 3.6.0 and install edgeR.

ADD REPLY • link 5.5 years ago nitesh.turaga ▴ 140

0

Entering edit mode

If you are able to use docker, and not worry about system dependencies,

In a Dockerfile

FROM bioconductor/bioconductor_full:RELEASE_3_9

RUN R -e "BiocManager::install('edgeR')

Then build,

docker build -t bioc-edgeR:latest -f <path>/Dockerfile

Use

docker run -it bioc-edgeR:latest R

ADD REPLY • link 5.5 years ago nitesh.turaga ▴ 140

0

Entering edit mode

The combination of R and limma versions you have will not run correcly, as has been noted here: https://stat.ethz.ch/pipermail/bioc-devel/2019-June/015240.html

You could either install R 3.6.0, or else down-grade limma and edgeR to Bioconductor 3.8 versions designed for R 3.5.X. In either case, everything would work correctly, including rpkmByGroup.

I explictly declared limma's dependence on R 3.6.0 in version 3.40.1 but not in 3.40.0.

ADD REPLY • link 5.5 years ago Gordon Smyth 52k