[BioC] edgeR calcNormFactors NaN

[Mark Robinson]

Hi Marina.

A couple of things would help me understand your problem:

-- can you give the output of your sessionInfo() ?  For one thing, it looks like you are using an older version of edgeR, so it would be worth trying the latest version.  I remember a fix we put in awhile back that accounted for lower-depth datasets that have a lot of ties in the log-ratios.

-- what does colSums(counts) and colSums(counts==0) give?

> On the other hand when I take a subset of the columns including the ones giving me f NaN v4 and v12 
> I do not obtain NaN anymore.  

I don't understand exactly what this means.  Certainly, if you take out columns V4 and V12, there shouldn't be any problem.

Alternatively, send me a data matrix off-line and I can take a look.

Cheers,
Mark




On 2011-01-29, at 12:01 AM, Marina Naval Sanchez wrote:

> Hi all,
> 
> 
> when doing calcNormFactors using edgeR I obtain some NaN values.
> 
> f=calcNormFactors(counts)
>       V2        V3        V4        V5        V6        V7        V8        V9 
> 1.0000000 0.9987462       NaN 0.7550607 0.6139919 0.8603117 0.8594895 0.7910109 
>      V10       V11       V12       V13       V14       V15       V16       V17 
> 0.7967414 0.7837294       NaN 0.8136299 0.8136299 0.8136299 0.8136299 0.8679718 
>      V18       V19       V20       V21 
> 0.8575024 0.8575024 0.8575024 0.8575024 


> I would like to know what is producing this effect. I tried to look at the calcNormFactors but I do not know what calcFactorWeightted does.


> function (dataMatrix, refColumn = 1, logratioTrim = 0.3, sumTrim = 0.05, 
>    doWeighting = TRUE, Acutoff = -1e+10) 
> {
>    if (!is.matrix(dataMatrix)) 
>        stop("'dataMatrix' needs to be a matrix")
>    if (refColumn > ncol(dataMatrix)) 
>        stop("Invalid 'refColumn' argument")
>    apply(dataMatrix, 2, .calcFactorWeighted, ref = dataMatrix[, 
>        refColumn], logratioTrim = logratioTrim, sumTrim = sumTrim, 
>        doWeighting = doWeighting, Acutoff = Acutoff)
> }
> 
> 
> Thank you very much.
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at r-project.org
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor

------------------------------
Mark Robinson, PhD (Melb)
Epigenetics Laboratory, Garvan
Bioinformatics Division, WEHI
e: mrobinson at wehi.edu.au
e: m.robinson at garvan.org.au
p: +61 (0)3 9345 2628
f: +61 (0)3 9347 0852
------------------------------


______________________________________________________________________
The information in this email is confidential and intend...{{dropped:6}}