[BioC] DESeq2 stuck running

[Michael Love]

It's going to be in v1.4.5, I forgot to bump the version number.

Mike

On Sat, May 10, 2014 at 11:43 AM, Michael Love
<michaelisaiahlove at gmail.com> wrote:
> hi Mingfeng,
>
> Thanks for reporting this and sending the example. I found the bug
> that produced this hang, and just pushed v1.4.4 (available in 1 day on
> Bioc), which fixes this:
>
>> library(DESeq2)
>> load("~/Downloads/tmpFile.Rdata")
>> system.time({nbinomWaldTest(ddsMF)})
>    user  system elapsed
>  14.841   0.777  15.634
>
> # or the whole pipeline:
>
>> system.time({DESeq(ddsMF)})
> using pre-existing normalization factors
> estimating dispersions
> you had estimated dispersions, replacing these
> gene-wise dispersion estimates
> mean-dispersion relationship
> final dispersion estimates
> fitting model and testing
>    user  system elapsed
>  38.768   1.457  40.240
>
> Mike
>
>
> On Fri, May 9, 2014 at 5:38 PM, Mingfeng Li <mingfeng.li at yale.edu> wrote:
>> Hi Mike,
>>
>> Thanks your so fast response. I would like to send you one example in Rdata.
>> You can get the information of the design, the normalization factor matrix.
>> please reproduce the following codes:
>>  library("DESeq2")
>> load("tmpFile.Rdata")
>> ddsMF = nbinomWaldTest(ddsMF);
>>
>>
>> Best,
>> Mingfeng
>>
>>
>>
>>
>> On Fri, May 9, 2014 at 4:12 PM, Michael Love <michaelisaiahlove at gmail.com>
>> wrote:
>>>
>>> hi Mingfeng,
>>>
>>> With 6 samples, this is typically seconds to run. If it's taking longer
>>> than a few minutes, usually this means something is wrong. Could you make a
>>> small reproducible example and send me data and a script? If this is not
>>> possible, I'd need some more information, like the code you are using, the
>>> design, the normalization factor matrix, etc.
>>>
>>> Mike
>>>
>>>
>>> On Fri, May 9, 2014 at 3:14 PM, Mingfeng [guest] <guest at bioconductor.org>
>>> wrote:
>>>>
>>>>
>>>> It is the exciting news of the coming of DESeq2.
>>>> There are many updates and incorporation with other software, e.g., Wald
>>>> test and cqn.
>>>> My question is the nbinomWaldTest may stuck running for long time when
>>>> using multiple factors, e.g., running several days but no results or
>>>> messages coming out. It works well when I remove the "type" factor.
>>>> I paste the meta data in the following:
>>>>            condition type
>>>> S103         W  U
>>>> S149         W Y
>>>> S114         W Y
>>>> S98          K  U
>>>> S92          K  U
>>>> S159         K Y
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> Thanks.
>>>>
>>>>
>>>>
>>>>
>>>>  -- output of sessionInfo():
>>>>
>>>> > sessionInfo()
>>>> R version 3.1.0 (2014-04-10)
>>>> Platform: x86_64-unknown-linux-gnu (64-bit)
>>>>
>>>> locale:
>>>> [1] C
>>>>
>>>> attached base packages:
>>>> [1] splines   parallel  stats     graphics  grDevices utils     datasets
>>>> [8] methods   base
>>>>
>>>> other attached packages:
>>>>  [1] cqn_1.10.0              quantreg_5.05           SparseM_1.03
>>>>  [4] preprocessCore_1.26.0   nor1mix_1.1-4           mclust_4.3
>>>>  [7] DESeq2_1.4.1            RcppArmadillo_0.4.300.0 Rcpp_0.11.1
>>>> [10] GenomicRanges_1.16.2    GenomeInfoDb_1.0.2      IRanges_1.22.4
>>>> [13] BiocGenerics_0.10.0
>>>>
>>>> loaded via a namespace (and not attached):
>>>>  [1] AnnotationDbi_1.26.0 Biobase_2.24.0       DBI_0.2-7
>>>>  [4] RColorBrewer_1.0-5   RSQLite_0.11.4       XML_3.98-1.1
>>>>  [7] XVector_0.4.0        annotate_1.42.0      genefilter_1.46.0
>>>> [10] geneplotter_1.42.0   grid_3.1.0           lattice_0.20-29
>>>> [13] locfit_1.5-9.1       stats4_3.1.0         survival_2.37-7
>>>> [16] xtable_1.7-3
>>>>
>>>> --
>>>> Sent via the guest posting facility at bioconductor.org.
>>>
>>>
>>
>>
>>
>> --
>> *********************************************
>> Mingfeng Li, Ph.D.
>> Postdoctoral Associate
>> Department  of Neurobiology
>> Yale University School of Medicine
>> 333 Cedar Street, SHM C-327C
>> New Haven, CT 06510
>>
>> E-mail: mingfeng.li at yale.edu
>> Lab: (203) 785-5941
>> Lab website: www.sestanlab.org
>> *********************************************