Housekeeping genes validation - (Sep/22/2008 )

Hi everybody,

I'm checking if the 18s is working as an hk gene in my samples, so I've run two controls and two treated samples (each in triplicates) and compared the Ct.
to better explain:
animal1: ctr1 v/s treated1 -> ∆Ct=1.4
animal2: ctr2 v/s treated2 -> ∆Ct=-1.3

(animal 1 and 2 are different individuals of the same species treated in the same way)

Now the questions:
Is that right that the ∆Ct between ctr and treated must be <1 to assume that the gene is an hk? Can I go away with a simle comparison of the Ct averages or do I need standard deviation and all that sort of stuff?

thanks a lot!

Raffaella

Morning!

First, the Delta Ct is defined as the difference between two samples (Ct1 - Ct2) not the ratio.
If you have a good reference gene the ratio between those two samples should be one (Ct1 / Ct2 = 1) (see Vandesompele et al 2002).
But how do you then control for different reverse trascription efficencies in you different samples? This might mask a real difference or act as a difference that is not existing.
I just tried the Vandesompele method to validate 8 HKGs because it is independent of the RT efficency in different samples. But as I repeated the measurement for each sample it turned out that I measured complete crap. So I strongly advice you not only to make replicates within the real-time PCR but also to make replicates of each sample (RNA extraction + reverse transcription) to see if you are measuring the reality (not like in my case...).
And to be really, really sure you should also increase the number of samples and do all that sort of stuff like averages, standard deviation...
An N = 2 is not very reviewer convincing ;-)

good luck!

Jan

I just got back to start....

Hi Jan ,

thanks for your reply..
I know that the ∆Ct is a difference and not a ratio.. so, ideally the ∆Ct between 2 samples should be close to 0 and the ratio 1. But what's (if there is one) an acceptable range for the ∆Ct or the ratio to accept that gene as an HKG?
and about the low number of samples... this is a sort of pilot-study...
Cheers
Raffaella

Hi,

Sorry, you're right, I should read more carefully.
I think there is no defined threshold value for this ∆Ct. Most of the times people compare several genes against each other and take the most stable ones. I can only think of the Pfaffl method, where the reference genes are supposed to have a standard deviation <1. So you can increase your number of samples and have a look how variable your ∆Ct is (so you need SD and that stuff...) and if it is very low and constant you can decide "ok, I'll take it". But as far as I know there is no threshold value like "ok, if it's 0,5 we accept it, if it's 0,6 we won't". Maybe there is, but I just don't know. I would suggest to increase your sample number to get a feeling for your data and see how much variation you find in your ∆Cts and also to analyse the replicates to see how much of this variation is an artefact of run variability.

all the best
Jan