[BioC] illegal operand when I run DEXSeq

Martin Morgan mtmorgan at fhcrc.org
Thu Feb 28 03:16:07 CET 2013


On 2/27/2013 4:48 PM, Steve Lianoglou wrote:
> Hi,
>
> I just wanted to ensure that you were using the latest version of the
> DEXSeq package, which you are.
>
> I'm not sure what other help I can offer other than the suggestion
> that it seems like it's time to do a bit more in depth debugging on
> your part, which won't be too easy if you're not comfortable with
> doing these types of things.
>

This

 > ecs <- estimateDispersions(ecs)

  *** caught illegal operation ***
address 0x2b101e337220, cause 'illegal operand'

is a bug at the C level. Here...

Traceback:
  1: drop(exp(X %*% beta + offset))
  2: glmnb.fit(mm, y, initialGuess, mf$offset, coef.start = start)
  3: fitted.values(glmnb.fit(mm, y, initialGuess, mf$offset, coef.start = start))
  ...

suggests that things are going wrong in drop(), but it could be that the error 
occurs before this. I looked at the C code in drop() for likely candidates, but 
did not have success. You might gain some hints with

   trace(drop, quote(print(str(x))))
   ecs <- estimateDispersions(ecs)

which will print out an abbreviated version of the argument that drop() sees; 
perhaps it is unusual in some way? On linux you can try debugging C code, as in

   http://bioconductor.org/developers/c-debugging/

The 'valgrind' section is probably less intimidating that gdb. it might be 
easier to create as minimal a subset of 'ecs' as possible that still reproduces 
the example, and share that either with the list or more narrowly.

Hope that helps,

Martin

>>From your original output, it seems like the problem must be somewhere
> in the top 100 genes being tested, since there wasn't a "." printed to
> the console (I think it prints that every ~ 100 genes to keep you
> informed of it's progress).
>
> Perhaps you can see if it works when you exclude the top 100 genes,
> and debug backwards from there?
>
> Digging into the estimateDispersions function and having it print to
> the console the name of each gene as it is estimating dispersions
> would be another idea.
>
> All of this is assuming that your input data is sane, which we have no
> way to verify from our end. Are the `counts()` of your ExonCountSet
> sane, for starters?
>
> -steve
>
> On Wed, Feb 27, 2013 at 4:40 PM, Ou, Jianhong <Jianhong.Ou at umassmed.edu> wrote:
>> Hi Steve,
>>
>> Because R is crashed after I run the code, I can only show you my sessionInfo before that.
>>
>>> sessionInfo()
>> R version 2.15.2 (2012-10-26)
>> 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] DEXSeq_1.4.0        Biobase_2.18.0      BiocGenerics_0.4.0
>> [4] BiocInstaller_1.8.3
>>
>> loaded via a namespace (and not attached):
>> [1] biomaRt_2.14.0 hwriter_1.3    RCurl_1.95-3   statmod_1.4.16 stringr_0.6.2
>> [6] tools_2.15.2   XML_3.95-0.1
>>
>>
>> Yours sincerely,
>>
>> Jianhong Ou
>>
>> jianhong.ou at umassmed.edu
>>
>>
>> On Feb 27, 2013, at 4:32 PM, Steve Lianoglou wrote:
>>
>>> Hi,
>>>
>>> On Wed, Feb 27, 2013 at 4:04 PM, Ou, Jianhong <Jianhong.Ou at umassmed.edu> wrote:
>>>> Hi All,
>>>>
>>>> I got illegal operation when I run DEXSeq.
>>>>
>>>> May I know the possible reasons for this error?
>>>
>>> Can you show the output of `sessionInfo()` after you load DEXSeq so we
>>> can see the package versions you are using?
>>>
>>> (also, please be more prudent in the size of the "log" you paste into
>>> the email you send to the list ... having a list of all of the genes
>>> in your study is a bit much ;-)
>>>
>>> -steve
>>> --
>>> Steve Lianoglou
>>> Graduate Student: Computational Systems Biology
>>> | Memorial Sloan-Kettering Cancer Center
>>> | Weill Medical College of Cornell University
>>> Contact Info: http://cbio.mskcc.org/~lianos/contact
>>
>
>
>


-- 
Dr. Martin Morgan, PhD
Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N.
PO Box 19024 Seattle, WA 98109



More information about the Bioconductor mailing list