Essentially, I simply want to produce a principle components analysis using qPCR data, to look at the similarities in gene expression across 10 varieties. I found HTqPCR, though it asks for ct values and has its own normalisation routines... do I need to use these normalisations if I already have the data normalised using a housekeeping gene and control samples?
I am not a statistics professional, but when I read Normalization, it rings a bell, see two papers which put in doubt the normalization view:
Limpert, E., & Stahel, W. A. (2011). Problems with Using the Normal Distribution – and Ways to Improve Quality and Efficiency of Data Analysis. PLoS ONE, 6(7), pp. e21403. http://dx.doi.org/10.1371%2Fjournal.pone.0021403 AND http://www.ask-force.org/web/LogNormal/Limpricht-Problems-Using-Normal-Distribution-2010.pdf
Limpert, E., Stahel, W. A., & Abbt, M. (2001). Log-normal distributions across the sciences: Keys and clues. Bioscience, 51(5), pp. 341-352. ://000169293900007 and http://www.facultyof1000.com/article/nonpub48840/evaluation and http://www.botanischergarten.ch/LogNormal/Limpert-Bioscience.pdf
I only know that a lot of statistics methods are still wrongly based on normalization, and if you ask a professional statistitian, he would agree. Specifically about your question I am unable to give an answer, sorry. Klaus
Thanks for your comments, they are helpful to think about it. Seems to me, that it makes sense to use internal standards (i.e. housekeeping gene) to normalise data rather than a statistical computation.
- Badges
- Science method