cleanUpPhenoData in arrayQualitymetrics disregards the intgroup columns for phenodata structures with many columns.

This line (https://github.com/Bioconductor-mirror/arrayQualityMetrics/blob/master/R/prepdata.r#L245) does not take into account intgroup columns; it might if they have low enough indices though:

wh = wh[seq_len(maxcol)]

One possible solution would be sth like the following, added on line 215, for example (https://github.com/Bioconductor-mirror/arrayQualityMetrics/blob/release-3.4/R/prepdata.r#L215):

newpd <- pd[, intgroup]
others <- setdiff(names(pd), intgroup)
pd <- cbind(newpd, pd[, others])

Dear Flo

Thank you for the report and precise dissection of the problem.  This should be fixed in arrayQualityMetrics version 3.31.2 in the devel branch. (Thanks to Andrzej Oles for testing and completing my fixes!)

Wolfgang