Dear,

I'm having a problem with the ttestCtData function from the HTqPCR package, which I use to analyze my BioMark Fluidigm data.

In most cases there is no problem:

> qDE.ttest.Nanog <- ttestCtData(q.norm[,c(1:9)], groups = conditions[c(1:9)], calibrator = "R-L_KD", stringent = FALSE)
> head(qDE.ttest.Nanog, n=2)
     genes feature.pos     t.test      p.value adj.p.value       ddCt        FC meanCalibrator meanTarget categoryCalibrator categoryTarget
40   Nanog   feature30 -16.290155 2.477252e-05 0.002204754  1.2277206 0.4269915       15.50882   16.73654                 OK             OK
84 Zcchc12   feature71   6.271628 4.370851e-04 0.019450286 -0.5513761 1.4654829       16.03707   15.48569                 OK             OK

However sometimes I get this error

> write.table(qDE.ttest.Nanog, file = "/Users/ruben/Dropbox/Data/qPCR/results/BioMark/Fluidigm4/ND2_fluid4/Ttest/ND2_ttest_Nanog.txt")
> qDE.ttest.Rest <- ttestCtData(q.norm[,c(1:5,10:13)], groups = conditions[c(1:5,10:13)], calibrator = "R-L_KD", stringent = FALSE)
Error in t.test.default(x[, g1], x[, g2], alternative = alternative, paired = paired,  : 
  data are essentially constant

However when I change the normalization from quantile normalization (q.norm) to for example norm.rankinvariant (nr.norm) this error doesn't occur anymore.

All of the samples are different, I compare 4 biological target replicates to 4 from the control. Could it be due to the quantile normalization? And would it be ok if I used the norm.rankinvariant normalization if I encounter this error?

Best regards,

Ruben
	[[alternative HTML version deleted]]