Hi again. I just want to stress that it is correct to say that ct values of PARTICULAR SAMPLES should ideally be within 0.5 cycles between technical replicates within a given assay and also between assays; that is the expression is consistent high level and reproducible. Further more in general you would want you housekeeper to be constantly expressed in all samples of a given type notwithstanding small differences due to sample purity and amount. That is to say that small differences BETWEEN BIOLOGICAL REPLICATES due to differences in sample quality and/or quantity are acceptable - indeed even to be expected - which is the necessity for data normalisation - but differences of more than 0.5 cycles BETWEEN TECHNICAL REPLICATES; THAT IS THE SANE TYPE OF SAMPLE AND SAME SOURCE OF SAMPLE ARE UNACCEPTABLE.  That said differences of day 5 or more cycles would imply something seriously awry like quality issues with samples due to preparation problems which would invalidate your assay

To reiterate you will encounter small differences between different RT samples due to small and inevitable differences in sample quality. This is the reason that good experimental design demands that you treat you amplify your housekeeping gene from the same sample as your gene of interest and normalise the latter with the former. For every type/ source of sample that is biological replicate you should amplify with both your housekeeping gene and gene of interest and normalise data from particular biological replicates with housekeeping genes amplified from the same biological replicate; ergo, you cannot just amplify a housekeeper from a particular biological replicate and use that to normalise gene of interest data from all biological replicates precisely because small differences between biological replicates in amplification efficiciency ( if both housekeeper and gene of interest) will inevitably arise due to small differences in purity ( of starting RNA and thus efficiency of cDNA synthesis) between biological repkicates

Take home message: technical replicates MUST be within 0.5 cycles in terms of Ct within and between cycles.  Biological replicates however might differ by 1-2 cycles due to purity differences which is why both housekeeping and GOI data must be geberated for each set of biological replicates for each assay

Ideally it should be, but practically many times it is not. But it should not considerably affect your results as you will normalize using the reference gene. Good luck. 

Hello Muhammed. What matters is not the the control gene is exactly the same in two different samples but rather consistent between technical replicates within an assay and between assays. RT Samples vary in quality as indicated by 260/280 ratios and 260/230 ratios. Consequently they vary in the efficiency with which they amplify and therefore for different samples even for the same gene you can get slight differences in Ct values. These purity issues also contribute to the ct value of the gene of interest. The question then becomes are differences between treated and control samples due to copy numbers differences in the gene of interest I.e differences in expression or due to differences in sample quality culminating in amplification differences. That dilemma can be corrected for by taking the ct value for the gene of interest away from the housekeeping gene. If a sample for example is highly pure and amplifies with gene of interest primers unusually well compared to the apposing sample then that will also be true for another gene e.g a housekeeper. Thus by subtracting the housekeeper from the gene of interest you correct for this inherent difference between samples - which has nothing to do with gene copy number - and thus normalise differences between samples. This corrected or delta ct value is then subtracted from the analogous delta ct value of the other sample and it is this delta delta ct value that represents just differences in expression and not artefacts 

In short slight ct differences are to be expected between different samples and these differences are used to correct or normalise data allowing an intrinsic calculation of expression by delta ct be made

Thank you dear Hall.

Regards

Yes, ideally the dCt should be as close to zero as possible. A 1 Ct difference is already a 2 fold difference in expression, so I would look for something with a 0.5 Ct difference or lower.

Hi again. I just want to stress that it is correct to say that ct values of PARTICULAR SAMPLES should ideally be within 0.5 cycles between technical replicates within a given assay and also between assays; that is the expression is consistent high level and reproducible. Further more in general you would want you housekeeper to be constantly expressed in all samples of a given type notwithstanding small differences due to sample purity and amount. That is to say that small differences BETWEEN BIOLOGICAL REPLICATES due to differences in sample quality and/or quantity are acceptable - indeed even to be expected - which is the necessity for data normalisation - but differences of more than 0.5 cycles BETWEEN TECHNICAL REPLICATES; THAT IS THE SANE TYPE OF SAMPLE AND SAME SOURCE OF SAMPLE ARE UNACCEPTABLE.  That said differences of day 5 or more cycles would imply something seriously awry like quality issues with samples due to preparation problems which would invalidate your assay

To reiterate you will encounter small differences between different RT samples due to small and inevitable differences in sample quality. This is the reason that good experimental design demands that you treat you amplify your housekeeping gene from the same sample as your gene of interest and normalise the latter with the former. For every type/ source of sample that is biological replicate you should amplify with both your housekeeping gene and gene of interest and normalise data from particular biological replicates with housekeeping genes amplified from the same biological replicate; ergo, you cannot just amplify a housekeeper from a particular biological replicate and use that to normalise gene of interest data from all biological replicates precisely because small differences between biological replicates in amplification efficiciency ( if both housekeeper and gene of interest) will inevitably arise due to small differences in purity ( of starting RNA and thus efficiency of cDNA synthesis) between biological repkicates

Take home message: technical replicates MUST be within 0.5 cycles in terms of Ct within and between cycles.  Biological replicates however might differ by 1-2 cycles due to purity differences which is why both housekeeping and GOI data must be geberated for each set of biological replicates for each assay

In addition to what has already been written; If your housekeeping gene Ct value is constantly and reliably different between your control and treated samples, it could mean that the difference is related to the housekeeping gene transcription actually being affected by the treatment. And this would also affect the endresults of your GOI; I guess the extend to which it's affected depends on how much your GOI actually change (a huge difference in GOI bc of the treatment could make a small influence on housekeeping gene less important). There could also be differences between control and treated samples caused by sample quality, as discussed above, let's say you are comparing highly inflamed to treated and therefore healthy tissue, or the sampling conditions were different etc.

Laurence Stuart Dawkins-Hall Thank you for your in depth answer. I too am having trouble getting consistent results in housekeeping genes in mouse embryos. I am getting differences of about 3-4 cycles between biological replicates, but my triplicates are always within 0.5 cycles difference. I have tried 13 housekeeping genes and none of them are giving me < 3.5 cycles difference between biological replicates. How do you determine if this is 'good enough' or we need to keep searching? There seem to be no published guidelines on this.

Dear Heather

• It is surprising that a given housekeeping gene is yielding < 3.5 cycles in difference
• I am assuming you are talking about the same housekeeping gene primer set by amplified from different cDNA samples; as opposed to the Ct values between housekeeping genes ?
• If as I suspect the former then take a look at your RNA and in these circumstances make sure you make cDNA from 1-2ug of total RNA which should have an OD 260/280 ratio consistently of 1.8 to 2.2 and a 260/230 ratio of > 1.0 and perform RT for 1-2 hours with random primers
• In addition, make sure you use an annealing temp of Tm-3 to Tm-5C (in these circumstances) but verify you obtain one strong specific band by agarose gel under such 'relaxed' conditions
• Then take that annealing temp into your qPCR reaction as your second annealing/extension temp PROVIDING that temp does not fall below 58C
• Irrespective, if the above does not lead to more concordant data then ultimately what matters is your normalised Ct value; That is the delta Ct value of your Treated/control sample minus your housekeeping Ct: The rational for this - which is obviously the reason for normalising - is that even if a sample Ct yields a Ct value of (say) 3.5 cycles < than another sample (biological replicate) then the Ct value of the housekeeper is also likely to be shifted accordingly, i.e. < 3.5 cycles
• In conclusion if these normalised values are within 1-2 cycles and if your Ct values are within dynamic range then your data will still be valid
• To determine if your samples are within dynamic range, perform a serial dilution of your cDNA and then plot that against the log Ct: This should produce a flat line as explained in the word document attached
• I hope that helps some...