GeNorm has indeed been phased out and is available as much improved genormPLUS module in Biogazelle's qbasePLUS software. The formulas for the identification of stably expressed reference genes are in Vandesompele et al., Genome Biology, 2002. If you already have 2 good reference genes, then you need to calculate the arithmetic average of the Cq values.

Thanks ))

One of my favorite quotes of J.V. (who posted above), is from his paper mentioned:

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC126239/

Especially commit to memory the sentence in that paper:

"Taken together, our data and these studies clearly show that ideal and universal control gene do not exist. This warrants the search for stably expressed genes in each experimental system, and for the development of an accurate normalization strategy."

This sentence reads like a fine strong coffee.

Over which the famous fable was written:

Santa Claus, the Easter Bunny and the 10 best reference genes in the world (for every sample type and every experimental treatment system) show up at a party in Manhattan, New York. Bunny turns to Claus and utters: "Why Mr. Claus, who do you think those 10 non-existent gents are?" At which point Claus turns to the ephemeral lagomorph and bellows: "Well my fine furry fellow, I haven't a clue, but perhaps genormPLUS and the elves can conjur some sense into them." Next, and with immediate horror, the 10 reference genes run screaming from the party at the 2 strange voices coming out of nowhere! They all then quickly go shopping for new pairs of jeans, singing "genormPLUS, genormPLUS, genorm all the way..." ThEnd. [Author: Some non-existent fellow].

Moral: Agreed, genormPLUS is the best option for this given the excellent (and extant, I might add) scientists behind it.

I see the issue was cost in the original post...

That indeed is a catch.

In that case... you would carefully identify (for your treatment system and sample type) what may be the top 10 or 20 candidates for ref. genes (Google away). Then, during qPCR, you would load the exact same amount of sample per reaction (after you have proven what is a good amount to add [not too little, not too much]) - but being absolutely certain you have added the exact same amount to all. Then you would want to see the same Cq value in all samples (with treatment and without) for a good 'stable' reference gene.

Since any gene, theoretically, could be used as a reference gene (if it doesn't change its expression after treatment), you have many to choose from - including ones that have never been explored...

So, short answer: look for 2 (or ideally, 3) genes that give you the same Cq +/- 0.2 Cq units in all samples, and take their arithmetic mean and proceed to normalize. But do not use the 2 to the -delta delta Cq equation ever. Always use the efficiency-corrected Pfaffl equation to get your final ref-gene-normalized fold change values. The Pfaffl equation is the more evolved form of the 2 to the -delta delta Cq equation as it takes efficiency into account.

If you should experience a windfall from a near or distant relative, genormPLUS is worth the expense.