I'm using the R-package lumi for the analyses of Illuminas HT-12 v4 expression chips. I currently have a problem with the probe detection. I want to use the mean of the negative control beads for probe detection. More precisely I consider a probe as detected which has a higher expression then 10% over the mean of the negative control beads. My problem is, which expression values should I use as template? I read that one should use the already background corrected expression values but I'm not 100% sure about it. What is your suggestion. Should I use the background corrected or the raw data? I mean the negative beads represent the background noise and if the raw expression values are higher then this (or my threshold: 10% over the mean of the negative control beads) then the transcript should be considered to be expressed. Please correct me if I'm wrong. Maybe I have a error in reasoning.
I used Model-Based Background Correction for my data.