Dear all,

Does fluidgm qt-pcr data which has normalized against a reference gene (eg GAPDH) on each plate

have to be normalized further within and between plates. If so, How does one know which method

to use (quantile, cyclic loess) etc. The data is to be analyzed with Limma.

Thanks and best wishes,

Rich

Richard Friedman, PhD

Columbia University Cancer Center

The reason why I use cyclic loess for PCR data is so that the CT values don't have to pre-normalized against a reference gene. Instead one uses the reference gene as an up-weighted control gene in the cyclic loess normalization, see

  qRT-PCR Sample Maximization software analysis and proper experimental design advice or

  DE analysis of PCR array [was: dataset dim for siggenes]

and this gives a useful compromise between reference-gene normalization and global normalization.

If you only have pre-normalized data, perhaps you could quantile normalize, but are you sure you need further normalization at all?