[BioC] Limma missing values

kfbargad@ehu.es kfbargad at ehu.es
Wed Feb 8 17:19:13 CET 2006


Dear users,

I am trying to use limma for analysis of Q-PCR data.

I have a matrix of log2 intensities for 96genes X 20 samples that I 
have imported using read.table. Some of the cells (not many) have 
missing values because the PCR did not work for that gene on that 
sample and I left them blank. 
> myEset
Expression Set (exprSet) with 
        96 genes
        19 samples
                 phenoData object with 1 variables and 19 cases
         varLabels
                cov1: read from file

I read in the help vignette that lmFit can handle missing values, but 
when I try it, I get the following error:

treatments = factor(c(0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1),labels = c
("BD","C"))
> design = model.matrix(~0+treatments)
> fit <- lmFit(exprs(myEset),design)
Error in lm.fit(design, t(M)) : NA/NaN/Inf in foreign function call 
(arg 4)
In addition: Warning message:
NAs introduced by coercion

I checked the mailing archives and found a few emails saying that 
limma can handle missing values. I also tried setting na.rm=TRUE

Any ideas as to how to proceed? 

I also learned elsewhere that it is better to use a non-parametric 
test for PCR data. Any opinions about this?

Regards,

David



More information about the Bioconductor mailing list