[BioC] DESeq multifactor GLM Error attributed to zero counts (coded as NA) in dataset?

[Wolfgang Huber]

Dear Steve

I think the issue is fixed in the most recent version of DESeq, and will 
be dealt with even more gracefully in an immiment version. Simon will 
provide more specifics, on this list.

	Wolfgang

Aug/10/11 10:35 PM, Steve Vollmer scripsit::
> Dear listers,
>
>
> I get the following error message in the new (beta) version of DESeq when I
> try to implement the fitNbinomGLMs function [e.g.fitNbinomGLMs(cds, count ~
> treatment + condition)].
>
> Error in glm.fit(x = c(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0,  :
>    NA/NaN/Inf in foreign function call (arg 1)
>
>
> This error message occurs regardless of which method (pooled, blind,
> etc.) that I use to estimateDispersions.
>
>
> It seemed likely that the problem was due to NAs coded into the cds
> during the dispersion estimates, so I used a subset of the data where
> at least one count was observed in every sample for each gene
> transcript (this is RNAseq data). THEN, the error goes away.
>
>
> Has anyone figured out a sensible workaround to this problem?
>
>
> Cheers,
>
> Steve
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> 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


-- 


Wolfgang Huber
EMBL
http://www.embl.de/research/units/genome_biology/huber