I would doubt one of the the following cases:

1. If the cDNA has undergone repeated freeze-thaw cycle, it might have degraded to a small extent.

2. There might be any pipetting error.

3. The qPCR cyclers have different amplification efficiency in different wells of the block, try to place your samples in the same well.

Else, there is no reason to obtain such results

Dear Navjyoti Chakraborty

Thank you for your answer. We used new cDNA as we wanted to repeat our result with new lot of RNA samples.So no possibility of degradation by repeated thawing. There is less possibility of pipetting error as the triplicate was almost same(we used triplicate for each sample). And we also used the same wells as previous. Now what could be the possible reason and my question is, will it be problem for publication?

I assume you used the same master mix, instrument and operator. You state that you used new cDNA; what do you mean by that? You reverse transcribed the same RNA preparation? Since this step can be rather variable, this could be a source of your increased Cq value. You are carrying out a relative quantification experiment: what has happened to the Cqs of your target genes. Have they increased to a similar degree? If so, your relative changes should be similar.

The lesson is that you should carry out target and reference gene quantifications on the same samples and, if you need to reverse transcribe more cDNA (a good idea in any case to look at variability or when using biological replicates), you must re-quantify target as well reference targets. You may also want to add calibrators to be able to compare the results from different plates more direcytl.

Hi Tariqul Tareq,

How far apart did you perform the repeat experiments? There are a number of different reasons that you may have got different Cq values depending on this. Your enzymes (eg polymerase or reverse transcriptase) may have become less efficient over time, or the fluorescence signal of your reporter may have degraded slightly, both of which could account for a higher Cq value. Additionally, primer degradation may also be something to consider. Have you used any methods to determine your PCR efficiency? There are a number of ways you can do this - including using online software such as PCR Miner. (this may be useful if you have not run a standard curve for your primers recently, as you can just use your raw PCR data)

As far as I am aware, it is quite normal to get different Cq values for a reference gene analysis where there is some difference in time between running your repeats as there are a number of different variables that can change just as a factor of time itself, even if you are the most careful researcher. I have not worked much with plants, but I would assume there are similar difficulties experienced in working with cells, including slight differences in nutrients and growth conditions, and these factors may result in differences in transcription rate. Typically, it is total RNA that is quantified prior to cDNA synthesis, and this is not always the best indication of how much message there is.

I would be more concerned if you were finding variation within your runs in each experiment. The fact that there is little variation within each repeat is encouraging, and I would focus on this. Are you using any sort of program or algorithm to validate your choice, or just working off a delta-Cq?

As far as how it will affect publication, well that depends. Is your publication just focussing on these reference genes, or are you bringing in a gene of interest as well? I would be looking at variations in total transcription of all genes I had measured. If everything you are measuring has a change of 5-6 cycles in the new samples then your choice is probably valid, and you should be able to justify your choice fairly easily. But if transcription is more variable, then things start to get a little trickier.

Best of luck,

Annette.

Thank you everyone.

Dear Stephen Bustin,

yes i have used the same master mix, instrument and operators. New cDNA means, as we were trying to repeat our previous result(Biological duplicate), we isolated new total RNA and consequently cDNA synthesis as previously( same growth condition,same stress,same amount of RNA for cDNA synthesis as well as for real time PCR ). We are trying to validate reference genes under environmental stresses. So no target gene is our concern. We are working on just different common reference genes to validate which one is best to use for normalization purposes.

Dear Annette Jacobsen

Yes i know about PCR miner for primer efficiency check and is it reliable or acceptable by the reviewers? My publication focus is just on reference genes; to validate which one is more stable. We have used Norm finder, Gnorm and best keeper to analyze the result. They analyze the stability among numbers reference genes and that was quite OK; they calculated the best one. Now my concern is not the stability, as i am trying to repeat our previous result, i was expecting that cq values would be same as previous while working with same gene, but the reality was quite different. We have repeated one gene and it varied 4-5 cycles than the previous. If all the genes we are using varies from 4-5 cycles than the previous result, then will it be acceptable?Thanks for your detail answer

I still think it is a problem with your 'new' RT reactions: either the RT reagents (especially the RT enzyme) were compromised in some way, or, your second round of RNA isolations were more robust than your first round, and you ended up adding too much RNA to your RT reactions (thereby inhibiting the RT reactions differentially) giving you the weak and sporadic results you, as of most recent, have attained. Did you check the concentration (and quality; by RIQ/RIN analyses) prior to your 2nd round of cDNA syntheses? If everything were truly held identical to your first experiment(s), you should not be getting this discrepancy, and, in most cases with qPCR reliant on a preliminary RT step to create cDNA, the problem(s) lie(s) therein. RT in a lot of cases actually stands for where the "Real Trouble" with the subsequent qPCR is.

Hi Tariqul Tareq,

I have used PCR miner in the past, and have found it to be fairly reliable, and to correlate relatively well with standard curve efficiency values. I used both - standard curves were necessary as I was using newly designed primers so needed to evaluate linearity as well as efficiency, and I used miner because it was free and available and I was curious about it! I believe there are other programs that can do the same thing, although I have not used them. As to how reviewers view miner data, as yet I cannot say as our paper has not been submitted - ask me in a couple of months!

The three programs you have used, I have used also, so I am familiar with how they work. I am curious as to why you are running additional data points now? Is it just to double check your results, or do you intend on using these data for your reference gene study? If it is the former, then demonstrating that all gene expression measured changes by the same amount (4-5 cycles) may be enough. If this is the case, then it is more likely to indicate a global difference - either biological or experimental - rather than a specific change to expression of one or more genes. I agree with Jack Gallup that the RT reaction is a common source of this kind of problem, but still think there are other issues that should also be considered.

If it is the latter, and you are planning on using the new data with your old results, then you will need to try and find some way to compare them. Ideally, as Stephen Bustin suggested, you should run all samples for a qPCR analysis of a specific gene together in the same plate, and if you know this will not be possible, best practice is to use a calibrator so that different plates can be compared. If you have not included a calibrator and do not have the resources or time to rerun the whole experiment, then one thing I would try for starters is to run some of the old sample and some of the new sample together on the same plate - in that way you will see if the old Cq values can still be repeated, whether the results *can* be compared, and whether there are any issues with the PCR process (primers, enzymes, machine, etc). If you do not have any of the old sample left, it is going to make it very difficult to compare your data. This would be the case even if your Cq values were similar - the fact they are not makes it even more important.

As to whether the reviewers will accept this method - again, ask me in a couple of months! I recently had to deal with a similar problem as you, in that I had to add in additional data after the first study had been completed. In hindsight, I now think it would be a good plan to include a calibrator in every run when doing this kind of study - this would have saved me a lot of time and effort. :)

Best regards,

Annette.

Dear Annette

Thank you. Yes you are right. I am going to double check my result, so i needed to run additional data point. Now it is not possible to run previous sample as i dont have any of them left and we used Nano drop(Roche) which contains only 32 wells. So its not possible to run many samples. I have go through with numbers of paper regarding validation of reference genes and all of them showed a figure contained the cq value differences within a same gene under different stresses. The differences are quite same as ours'. I dont know whether they get the difference under single stress or not. I have attached a paper; see the figure 1, please can you explain what they have done? whether they get differences in same gene under same stress or among stresses? I have asked to some of the authors but they didn't reply yet.

Hi Tariqul,

OK, if you have no old sample left, then showing a consistent change in Cq across all genes is going to be your best option, save repeating the whole experiment. :)

That figure - I have prepared something similar for our draft paper. In my case we had five experimental conditions, and I used that kind of figure to show the Cq variability across all stressors. This makes the most sense to me - even if you do examine individual stressors later, it is nice to give an indication of total variation (prior to analysis) in all conditions. It appears to me these authors have done the same thing, and it is the change under all stressors measured. Hopefully you hear back from them soon!

Good luck,

Annette.

Thank you very much Annette Jacobsen.