# how to select a calibrator for relative gene expression in real-time PCR - also about the inconsistant about the triplicate of real-time PCR (Aug/25/2008 )

Hi, dear all,

I am running real-time PCR for the relative expression of proinflammatory cytokines in response to infection. I use ABI system SDS software and SYBR Green. The expriment design is like this: 0h (no infection), 3h (collect RNA 3h post infection), 6h (collect RNA 6h post infection). I have triplicate samples for every time point. so there are 9 sample 0h-1, 0h-2, 0h-3; 3h-1, 3h-2, 3h-3; 6h-1, 6h-2, 6h-3. During data analysis, I have to select one of the zero time point sample as the calibrator, that is to say I set the value of this sample as 1. I set 0h-1 as 1, and the value of 0h-2 is 0.7, the value of 0h-3 is 1.3.

I am wondering if I set 0h-2 as the calibrator, that is to say the value of 0h-2 is 1, so the relative expression value of all the sample will change and increase 1/0.7 times. Form the three zero point sample, which one should I select as the calibrator? are there any rules for calibrator selecting? why there are so big a difference for the triplicated zero time point sample which have not been infected?

Thanks in advance

considering the high variance of your (untreated) calibrator: i assume, that the cytokine expression of the uninfected cells is very low to near the detection limit (Cts around 34 or more). moreover, using this as calibrator would introduce a large error in your sample measurements, the higher the difference of the Cts between calibrator and sample the higher the increase in error. I would recommend to pool cDNA from different treatment stages so you get some kind of average expression level and use this as calibrator. you can even calculate the arithmetic mean of all sample Cts and take this as calibrator. The numbers may be different but the relative quantities are always the same.

BTW: you would take the arithmetic mean Ct of your three 0h samples as calibrator, not only one

Hi, we had sort of the same trouble when doing RT-qCR for analyzing citokine expression, and at that time we decided to avoid to use the calibrator and to calculate the 2^-ddCT; we worked with the ddCT instead, that way our results were more accurate since we didn't have to worry about std deviation between calibration sampled due to the high CT. Lately we are working with ELISA, it's far easier and even more accurate if you need to know whether a given citokine rises along time. Good luck

BTW: you would take the arithmetic mean Ct of your three 0h samples as calibrator, not only one

Hi Ned Land, Thank you very much for your kind suggestions. This is the first time I heard that use pool cDNA from different treatment stage as the calibrator in qPCR. However, I am wondering how to explain the relative expression of this comparison. Since what I want to know is to see if treatment such as infection can stimulate or decrease the expression of specific cytokine. so I compare the expression level of treated group with non-treated group (here I refer to as 0h), and get the relative expression level. So, what is the significance of compare the expression level of speific treated group with the pool cDNA? if the expression level of speific treated group is higher than the average expression level, maybe we can say that the treatment increase the expression. However, if the expression level of speific treated group is lower than the average expression level, if doesn't mean that the treatment decrease the expression.

I also think that it would be better to take the arithmetic mean Ct of the three 0h samples as calibrator, I want to know how to do it in the ABI SDS software. it seems that only one sample can be selected as the calibrator in SDS software.

Thanks again

Hi, I am bewildered by what you said that you worked with the ddCT instead. Could you give more detail of how you analyze the relative expression data. Thanks

can you provide some additional information? what are your Cts of the 0h, 3h, 6h time points?

and again, the choice of calibrator doesn't change your relative quantification results. its more a cosmetical operation.

IN the end it doesn't matter which sample you take as calibrator. the calibrator should just have the following characteristics:

-all genes of interest have to be expressed in the calibrator

-enough sample material of the calibrator is available throughout the whole study

So you can also mix an aliquot of all samples and use this as a calibrator.