Header Row In Write.Exprs Is File Names Instead Of Designated Sample Names

Question

[Gotgenes]

I am trying to obtain normalized expression values for Affymetrix microarray data that we have using bioconductor. The script I have is mostly there, however, I am irked that the row header describes the samples by the file names of the CEL files, rather than the sample names designated in Targets.txt.

Here is an excerpt of my Targets.txt file:

# file for use by limma and affylmGUI. the targets are per-condition and per-time-point.        
Name    FileName    Target
feh.rep1    FH1.CEL    feh
feh.rep2    FH2.CEL    feh
feh.rep3    FH3.CEL    feh
...

Here is the R script:

library(affy)
library(gcrma)
# phenotype data
pd <- read.AnnotatedDataFrame("Targets.txt", header=T)
affy.data <- ReadAffy(filenames=pd$FileName, phenoData=pd)
# gene expression data, normalized by GCRMA
eset <- gcrma(affy.data)
write.exprs(eset, "cs_hm_feh-expression-gcrma-2011-01-12.tsv", sep="\t", row)

I have tried using

affy.data <- ReadAffy(filenames=pd$FileName, phenoData=pd, sampleNames=row.names(pd))

however, that was unsuccessful. Further investigation shows that row.names isn't actually getting anything at all

> row.names(pd)
NULL

which I find perplexing, given that the object shows it has row names, which are exactly what I want (and expected) as my sample labels in the final CSV table:

> pd
An object of class "AnnotatedDataFrame"
  rowNames: feh.rep1, feh.rep2, ..., cs.8d.rep3  (27 total)
  varLabels and varMetadata description:
    FileName: 
    Target:

Any help is appreciated, as I wield R fairly ignorantly and can not figure my way through this one seemingly simple task.

Answer 1

[Brad Chapman]

Is this what you are looking for?

> sampleNames(pd)
[1] "feh.rep1" "feh.rep2" "feh.rep3"

I'm not an affy expert at all; a good R trick is to look up the documentation on the object with:

> ? AnnotatedDataFrame
[...]
 ‘sampleNames(<AnnotatedDataFrame>)’,
      ‘sampleNames(<AnnotatedDataFrame>)<-<ANY>’: Set and retrieve
      the sample names in ‘AnnotatedDataFrame’

[Gotgenes]

Thanks, Brad! Excellent suggestion reading the docs. I was reading the wrong ones (for ReadAffy and read.AnnotatedDataFrame). Worked perfectly using sampleNames=sampleNames(pd). I'm still surprised I had to manually specify this, but I probably made the wrong assumptions and/or abused the method calls.

[Brad Chapman]

Chris, agreed that it seems like it should work without manually specifying it; that's what made me so unsure I was answering your question correctly. Glad that it worked.