Multiplex gene expression normalization

Question

[yuggnurug]

This is probably a naive question but I will ask it anyway. In qPCR experiment, can you compare multiple target gene expression that have been normalized by using two different sets of housekeeping genes? One would ask why I would use two different sets of housekeeping genes. Well, the multiple target genes have widely different Cq values and I wanted to use housekeeping genes with similar range of expression for the those target genes. For univariate statistics it's no problem. But I am wondering how this would impact clustering analysis. Please enlighten me. :)

Answer 1

[Kevin Blighe]

It would obviously be preferable to have used the same housekeeper. Using different housekeepers, as you've mentioned, it's not so much an issue for univariate analysis (but still not ideal). Once you jump from univariate to multivariate analysis, though, you have issues because the relative quantifications are then immediately biased due to having used different housekeepers. It will affect the clustering of the samples. Depending on the differences in the range of the housekeepers that you've used, the level of bias will be either big or small. It may even be negligible (too small to be noticed).

I would just use a single housekeeper for everything.

You have not clearly stated the exact normalisation method, but, you could justify the use of multiple housekeepers in the way that you've done if you could additionaly compare their levels in a reference sample(s). This would be then akin to the ΔΔCt method. Here, you would compare the ratio of test gene to housekeeper gene in each sample and your healthy control, and then get the ratios of these ratios.

The only other situation where you could justify the use of multiple housekeepers in the way that you've done is if you were analysing data from multiple tissues, where a housekeeper in one tissue may have nil expression in the other tissue. Even then, some form of normalisation would be required afterward across multiple tissues.

Hope that this helps

Kevin