[BioC] Differences between Linux and Windows

[Wolfgang Huber]

Dear Ingrid,

would you be able to post your script together with the data needed to
run it? You can anonymize the data, or use one of the public datasets
http://www.bioconductor.org/download/experiments

This seems like something that needs to be chased up.

 Best wishes
 Wolfgang

------------------------------------------------------------------
Wolfgang Huber  EBI/EMBL  Cambridge UK  http://www.ebi.ac.uk/huber



> Hi
> 
> I have now made my script as short as possible and included set.seed(12345). The only differences between the Windows and Linux run are:
> 
> -	The path to the files.
> -	memory.limit(4000) used in Windows
> -	jpeg is used in Windows and bitmap is used in Linux.
> 
> I have also tried to use to different implementations of gcrma: justGCRMA() and gcrma(dataSet). The results are the same when I am using these to methods, but the results between Linux and Windows are still different: the number of differentially expressed genes is not the same, but most of the genes found are the same.
> 
> Regards,
> Ingrid
> 
> 
> 
>  
> Dear Ingrid,
> 
> This could be (among other things)
> (1) a floating-point round-off error together with the thresholds you
> use, or
> (2) a consequence of some randomization / subsampling in one the
> algorithms you use; for example, gcrma does that.
> 
> To diagnose this, you could
> (2) Use "set.seed(12345)" at the start of the script and see if you then
> get more identical results
> (1) wiggle your p-value cutoff (or whatever you do, you didn't say that)
> a bit
> 
>  Best wishes
>  Wolfgang
> 
> 
> 
> Ingrid H. G. Østensen wrote:
>> Hi
>>
>> I have run the same script on Linux and on Windows (sessionInfo are below) and I get different number of differentially expressed genes out in my VennDiagram. When I use Windows I get 53929 genes that do not belong anywhere, 667 in one group, 11 in the other and 68 that they have in common. When I run this on Linux I get 53930 genes that do not belong anywhere, 663 in one group, 14 in the other and 68 that they have in common. There is not much difference but it is a difference, any idea why? Both R/Bioconductors have been updated within the last 14 days or so.
>>
>>
>>> sessionInfo()
>> R version 2.4.1 (2006-12-18)
>> i686-pc-linux-gnu
>>
>> locale:
>> C
>>
>> attached base packages:
>> [1] "splines"   "tools"     "stats"     "graphics"  "grDevices" "utils"
>> [7] "datasets"  "methods"   "base"
>>
>> other attached packages:
>> hgu133plus2probe   hgu133plus2cdf      hgu133plus2          annaffy
>>         "1.14.0"         "1.14.0"         "1.14.0"          "1.6.1"
>>             KEGG               GO         annotate            gcrma
>>         "1.14.1"         "1.14.1"         "1.12.1"          "2.6.0"
>>      matchprobes           xtable     RColorBrewer     affyQCReport
>>          "1.6.0"          "1.4-3"          "0.2-3"         "1.12.0"
>>       simpleaffy       genefilter         survival             affy
>>          "2.8.0"         "1.12.0"           "2.30"         "1.12.2"
>>           affyio          Biobase            limma
>>          "1.2.0"         "1.12.2"          "2.9.8"
>>
>>
>>
>>> sessionInfo()
>> R version 2.4.1 (2006-12-18) 
>> i386-pc-mingw32 
>>
>> locale:
>> LC_COLLATE=Norwegian (Bokmål)_Norway.1252;LC_CTYPE=Norwegian (Bokmål)_Norway.1252;LC_MONETARY=Norwegian (Bokmål)_Norway.1252;LC_NUMERIC=C;LC_TIME=Norwegian (Bokmål)_Norway.1252
>>
>> attached base packages:
>> [1] "splines"   "tools"     "stats"     "graphics"  "grDevices" "utils"     "datasets"  "methods"   "base"     
>>
>> other attached packages:
>> hgu133plus2probe   hgu133plus2cdf      hgu133plus2          annaffy             KEGG               GO 
>>         "1.14.0"         "1.14.0"         "1.14.0"          "1.6.1"         "1.14.1"         "1.14.1" 
>>         annotate            gcrma      matchprobes           xtable     RColorBrewer     affyQCReport 
>>         "1.12.1"          "2.6.0"          "1.6.0"          "1.4-3"          "0.2-3"         "1.12.0" 
>>       simpleaffy       genefilter         survival             affy           affyio          Biobase 
>>          "2.8.0"         "1.12.0"           "2.30"         "1.12.2"          "1.2.0"         "1.12.2" 
>>            limma 
>>          "2.9.8" 
>>
>> Regards,
>> Ingrid