*Hi BioC list, ** **In package beadarray, when I do boxplot of raw data, do I really have to do log2 transformation according to the example code in the tutorial: > boxplot(as.data.frame(log2(exprs(BSData))), + las = 2, outline = FALSE, ylab = "log2(intensity)")? The reason I asked this question is because I find some negative intensity values, which should prevent the log2 transformation. Another question is that whether there is an option in beadarray to filter out some probes across samples whose Detection P value is significantly lower than a threshold. Thanks a lot for your suggestion! Alex *** [[alternative HTML version deleted]]

Dear Alex, To filter probes based on some cut-off of the detection p-value score, try the code below (note, this assumes the detection scores are available. In this example, I have selected those with scores greater than 0.99 in each sample - you may need to adjust this for your dataset). ind = apply(Detection(BSData)>0.99, 1, all) BSData.filt = BSData[ind,] As for your plot, negative intensities produce a warning after you log2 and will be missing from the boxplot, however you should still see a figure of the remaining data. Best wishes, Matt >*Hi BioC list, >** >**In package beadarray, when I do boxplot of raw data, >do I really have to do log2 transformation according >to the example code in the tutorial: >> boxplot(as.data.frame(log2(exprs(BSData))), >+ las = 2, outline = FALSE, ylab = "log2(intensity)")? > >The reason I asked this question is because I find some >negative intensity values, which should prevent the log2 >transformation. > >Another question is that whether there is an option in beadarray >to filter out some probes across samples whose Detection P >value is significantly lower than a threshold. > >Thanks a lot for your suggestion! > >Alex