Normalization based on housekeeping genes

Question

[elenichri]

Dear Biostars team,

I have a gene expression dataset which is produced by Illumina HTv14 microarray platform. I have performed log-transformation and quantile normalization between samples based on all genes of the array. Now my boss asked to see the absolute levels of a gene of interest, thus asked me to normalize based on a set of housekeeping genes. I know it is not the best strategy for normalization but I need to do that...and I have never done that before. I did some reading on it and found that I have to normalize based on the expressions of these genes. I am using R. Would a function like normalizeBetweenArrays {limma} or normalize.quantiles.use.targets {preprocessCore} be suitable,by setting as targets the expressions of the housekeeping genes? I tried it but I get totally different genes to be diff expr between the conditions than the ones I was getting using quantile normalization based on all genes of the array. Do you have any suggestions?

Thank you very much!

Eleni

Answer 1

[andrew.j.skelton73]

Normalising to housekeeper genes shouldn't really be done with HT12 arrays, as far as I'm aware. The recommended normalisation is VST, followed by RSN or Quantile.

Housekeeping genes are only noted here by Illumina in reference to Control Probes (which are mostly used for background correction).

A better suggestion might be to show the expression of the housekeeping genes, to reassure your boss. Here is an example of how to do that. These control probes should be used to gauge how successful the array was in hybridisation, staining, etc.

[elenichri]

Thank you so much for these two resources Andrew. They are really informative!