In my idea, it isn't correct way because every rat has different genome structure, and you'd better examine every group separately.

GOOD LUCK

Even though this should theoretically give you the same average over the group overall, it would make you lose information about the variability between members of the same group. To my knowledge, you cannot do a statistical test without this information.

Thanks so much for your help.

What if I repeat the procedure three time with theses pool samples and report my data with this SEM?

I already read one nature journal's article which use this protocol. for more infprmation please find this link

http://www.ncbi.nlm.nih.gov/pubmed/12610626

That would make it so the standard deviation reflects the variability in your qPCR procedure (technical replicates) instead of the variability between rats (biological replicates), and it would also mask any technical variability in your initial procedures. It will certainly lower your perceived variability, but it would not reflect your actual confidence in the resulting value.

The only reason I would suggest doing this is if you are limited in the amount of sample you can take, and the amount of RNA you can detect is on the lower boundary of your detection. Then pooling samples will allow you to detect the amount of RNA at the more accurate part of your detection range.

If you have at least 3 different cDNA samples then you go for statistical analysis mainly one way ANOVA as well as you are getting sample mean and SEM. You want to do statistical analysis then no need to pull all the samples of cDNA from the same groups in a single tube.

i want to know why are you pull all the samples of cDNA  from the same groups in a single tube?.

Dear Saurav Saha

As I mentioned before, I think this method can decrease some errors which frequently happens because of the  instruments or some non- purpose personal mistakes, and also can save money a lot

Perhaps you have misunderstood the methodology in the nature article?  I agree as above if you pool all samples you have no way to differentiate between samples and cannot perform analyses such as above a which compares within group vs between group variances with an F test. For clarification you could supply the article link you are mentioning? 

I agree with other people here - although in theory it will give you the same average, you could actually be introducing an error. Imagine you have 3 samples and you pool them and get a "mean" of x from them when you run the PCR. Now consider that the RQ values (for example) are 10, 12 and 156. If you run each sample individually you will see the error that you have and can chose to either re-run to see if there was anomaly with that sample, or exclude it if it is repeatedly different from all other samples in that group. If you pool all of them together, your mean will be higher and you will never know if you've introduced such an error.

Thanks you all  for sharing helpful opinion . I am aware of this important error you all mentioned above, but  Do you believe our current method which needs three time repetition and every time has a completely  different result is the best? Which way is better? pooling and not having S.E.M or evaluating every separate sample and ignoring differences you have seen? Real time PCR have not a constant result even in two consecutive days. some of these error are because of RNA preparation methods (concentrations and so on), and the others are because of reagents and instrument and so on and in most of the time you are not aware of them.

Dear Marina

please explain what is exactly my  miss- understanding at that nature article again.

In my opinion, pooling of samples will not help you as far as science is concerned. If you pool your samples of different mice of the same group, you can avoid the variance, hence reduced error, but doing so itself is a big error. If your treatment is effective, you will get almost similar trend in all the subjects of the same group, which means lower deviation and significant result. If you are getting high deviation then it indicates 1) either your treatment is not effective enough and the deviation in expression may be due to some factor(s) other than your treatment (environmental stress, etc.); or 2) your housekeeping gene/endogenous control gene expression is varying with the treatment. I would suggest you to try with more than one housekeeping gene. If you still continue to get similar result as you are getting now, then you need to conclude your experiment with whatever result you are getting as the genes under investigation are not affected by the treatment you have given.

Hope it may help you. All the best. Remember - Biological replicates are extremely important in fool-proof conclusion of your experiments.

You will get different results on different days in terms of absolute values. However, there are ways around this. You can use calibrator samples which you run on each plate. These samples normalise plates between runs. Otherwise, you will get the same trend in data, but not the same overall values. You could also consider using fold change as a way around this. A true result will/should give similar differences between samples even if the absolute values change.

So if samples 1 gives a value of 50 and sample 2 a value of 100 on day one, and then on day two they give you values of 90 and 180, you have the same difference between the two samples in terms of fold change. So, you can use that to compare between plates. However, the best way is to establish standard curves and use calibrator samples, or use 384 plates where you can obviously put more samples on the same plate.

Thanks you all. I think you convince me now. I will try your suggestions.

Best

Farzaneh