Hi there, 

I've got an Agilent expression dataset which I'm trying to analyse using limma. 

Here are the issues:

1. I cannot read.maimages(agilentdataset) as suggested in the user guide. That prompts the following error:

Error in readGenericHeader(fullname, columns = columns, sep = sep) : 
  Specified column headings not found in file
In addition: Warning message:
In readLines(con, n = 1) :
  incomplete final line found on 'GSE85426_raw_data_original2.txt'

2. I seem to be able to only add the annotation from a separate annotation file by merging the data.frames. However, then I'm stuck at n.sv and lmfit() which it won't let me do, because the data.frame isn't a suitable format.

n.sv = num.sv(norm_merged,mod,method="leek")
Error in diag(dims[2]) - mod %*% solve(t(mod) %*% mod) %*% t(mod) : 
  non-conformable arrays

3. I've successfully turned the data set into an ExpressionSet using

eset <- ExpressionSet(exprs, phenoData=pdata)

However, olgio::rma() won't work on this. If I normalise and bg correct it using limma's features, I get data frames which once again won't work with lmfit()

4. I've also tried library(coRNAi) to turn it into a working file for lmfit.

norm_merged_d2 <- df2lmFit(norm_merged)
Error in df2lmFit(norm_merged) : 
  no residuals, need to estimate main effects and update dataframe

All in all, I cannot seem to find a suitable workaround to normalize and subsequently combat and lmfit this dataset, let alone successfully add the annotation file. The annotation is however the least of my worries right now, since merging the data frames by row.names does work and hence could be done after analysis to find out the names of the top 1000 DE genes. 

Does anyone have a suitable workaround of this? I apologise for the long description. I've been trying stuff on this for 3 days now, so appreciate any and all hints/help/links. 

Dear annecomplete,

don't try too many packages -- the complete analysis can be done with limma.

Unfortunately, you do not give us your call to read.maimages() and we don't know which columns it cannot find in your Agilent files. At some point, Agilent introduced a reduced ("lean" IIRC) file format which could cause this error, depending on the type of file you are trying to read. So, please: show us your commands.

(I wouldn't be concerned about the warning; I remember seeing it without doing any harm.)

Once you can properly import your data, it's just a matter of following an example from the limma users's guide.

Cheers,

 - axel