Compare different qPCR runs - really need help (May/02/2010 )

Hello everybody, I am really new to PCR, qPCR so I need help of experts

I started to work with different samples and i look for the expression of different genes in these samples. Since I have new samples every week, I cannot accumulate them and do just 1 run for 1 gene on one plate for all my samples. Instead, whenever I have them, I run a housekeeping gene for each sample and I run my genes of interest.

So my question is, what are the rules for comparing the expression of a gene for samples done on different days (without changing a protocol or the machine)?

I have heard of inter-run calibrator but i am not sure I understood well what it was. I was told that the expression of this non-relevant sample on different plates tells you if your qPCR was ok and did not differ from other ones. If this is true the relative expression of this non-relevant sample will be the same between different runs and this will allow you to compare your true samples (assemble them on one graph).

So did i understand correctly what an inter-run calibrator is? What should this inter-run calibrator be? Can I take just a completely different sample (organ) and look for the expression of a completely different gene? or should the calibrator express the same genes that I am looking for?

(Someone in my lab told me that in order to be able to compare my different runs, I must re-run one of my samples with each new sample, but it is just impossible because my samples are small and if i do it i will compare maximum 3 samples. Is this person correct?)

Your answers will help me a lot

Thanks a lot

hi,

you have to run this inter-run calibrator (IRC) on every plate with the same targets as your samples. it does not have to be one of your samples (or a pool of it) but it would be not bad because then you are really testing the same RNA/cDNA.
there are two common problems with samples as IRCs: firstly, scarce sample material. this can be solved by pooling many sample cDNAs. the second problem: every gene that is detected in your samples must be present in your IRC for obvious reasons. This can be also solved with a pool of samples as IRC (in most cases).

another possibility would be to use some kind of commercial qPCR reference RNAs which are pools from cell lines or tissues. look at clontech for that (PN 636690). maybe this would be an option to you as you may have only access to a small number of your samples at the beginning of the experiment.

and run your IRC on every plate at least in quadruplicates.

Thanks a lot for the answer

But can you tell me the meaning of the IRC.

I am looking at the expression of cytokines in mouse pancreas. I know which or them I must find upregulated in some samples and the other ones not. My IRC, if well chosen (I could make a pool of my two types of samples) will show a relative expression of these cytokines. Then what should I look at? I should make sure these cytokines are expressed at the same level by IRC in all my runs?

Why is it important that my IRC expresses all the genes I am looking for (sorry this wasn't obvious for me )? Is it a positive control in addition to being a calibration tool?

Actually, as an IRC I was thinking of using mRNA extracted from the spleen (because then I would have tons of it), then running it on all my plates and make sure the expression of a gene X (non relevant to cytokines) is expressed at the same level on my plates. Can you tell me if it will be wrong doing this?

The meaning of the IRC is to calibrate for technical variances between different runs i.e. variances that occur by different plastics, different master mixes, different machine running conditions,... and you have to calibrate for those differences if you want to compare samples on different plates. The simple reason why every GOI must be present in the IRC is that if you encounter a GOI that is only expressed in your samples but not in the IRC you cannot calibrate for this. Another practice of calibrate for technical variances between different plates is simply to adjust the same baseline and treshold settings for the different plates but I would not recommend this practice, it presumes a big confidence in the reproducibility of your real time machine.

if you want to study for instance IL-1 expression and your samples are spreaded on two different plates you have to run the IRC for IL-1 on every plate.

i second that. you will have to use your gene of interest as IRC to be able to compare between plates!

All right, you guys are great, now i understand the purpose of the IRC expressing my genes of interest.


Can you tell me how the calibration is done mathematically?

For example if in my first experiment the relative expression of IL-1 in my sample is 0.5 and of my IRC is 0.8.
Two months later, my second experiment shows that the expression of IL-1 in my sample is 0.5, buy my IRC value is 0.6. What should i do next?

you just set the IRC value to 1 and then calculate the relative expression of your samples.
so your IL-1 expression would be 0.625 and 2 months later 0.83 but if you are calculating with the ddCt method you first have to normalize to your endogenous control which can be again on a separate plate but in this case you don't need an IRC.

so if you have plate A with 50 samples (including IRC) and IL-1 as GOI and plate B with the same 50 samples and your reference gene, for instance, GAPDH you would first normalize all your IL-1 samples with GAPDH (dCt) and then use your IRC as calibrator sample (-> ddCt).

With your plate C (IL-10) and D (GAPDH) two months later you are proceeding analogously. Then you can compare the samples from plate A and C, both are realtive quantities compared to the same sample (=IRC).

This was a very short explanation but I hope its clear.

tea-test on May 3 2010, 05:09 PM said:

Hi,

I'm having some similar issues with my current calculations. In my case I want to first run the IRC and then normalize my data to the control sample of my experiment. I have run the same IRC in every one of my 70 plates, for each of the genes I'm analysing. Could you please have a look at my workflow to see if it makes sense:

Step 1: Normalize the Cq values for each target gene in relation to a reference gene (or a combination of reference genes). In this case I would do (Cq of target gene - Cq reference gene) for each of my samples, including my IRC. This would be my DeltaCq.

Step 2: Inter-run calibration: Normalize the DeltaCq of my samples in relation to the DeltaCq of my IRC. This would be (DeltaCq sample - Delta Cq of IRC), which would result in a DeltaDeltaCq.

Does my calculation make sense until here?

Potential set 3: From here, I still would like to use an extra calibrator, which would be the control sample for my experiment. For example, I'm testing animals infected by a pathogen at different time points, and I want to compare their gene expression levels to non-infected animals. Would that be possible if I need to do an IRC as well?

Thank you very much for your help!