The key difference between the two methods lies in how they handle amplification efficiencies. The ΔΔCT method assumes equal efficiencies, whereas the E-Method accounts for the actual efficiencies of the reactions. If the reaction efficiency is close to 2, the results from both methods should be relatively similar. However, if there are differences in amplification efficiencies between the target gene and the reference gene or if the efficiencies are not close to 2, the two methods can yield different results.

In your case, the difference in the size of the results (1.57E-06 for basic and 0.706 for advanced) can be explained by the fact that the E-Method takes into account the actual amplification efficiencies, which may not be close to 2, leading to different numerical values compared to the ΔΔCT method. The E-Method is considered more precise because it considers these variations in efficiency, making it suitable for situations where accuracy is critical. I hope this will help.

Thank You for your answer.

The efficienciy for the target gene is 2,06 and efficiency for the reference gene is 2,116 (calculations for a standard curve based on dilutions of the tested sample), could the efficiency in the further examined samples differ from it?

What numbers are you actually comparing?

When you say "1.57E-6", what is this the result of?

I assume it can't just be ddCt, because Ct values are never calculated to 6 decimal places (and a ddCt of 1.57e-6 is basically "about zero", i.e. no change).

So it would be very helpful to know what you're measuring (how many replicates, how many samples, how many reference genes, etc) and what you're using from those measurments, and how, to derive the values you've quoted.

Thank You for your answer.

The values I provided are the Target/Ref values from the LC480 analyses.

I have three biological replications and three technical replications. There are no significant differences between replications.

I just noticed the differences between the analyses.

Another example (the same sample): 1,62E-6 the Target/Ref value from the Basic Relaive quantification; 0,6311 the Target/Ref value from the Advanced Relaive quantification

One target gene/ one selected reference gene (the best one for the experiment).

Additionally there are quite a big differences between the amout of the target and reference gene in the sample. The Cp of reference gene is within 8 and the Cp of the target gene is withn 30.

Ok, so...I still have no idea what your numbers mean.

Honestly, I would export the Cq data to excel and do it manually, because you'll have a lot more control over everything then.

First things first: a Cq of 8 is...really high expression: I am going to assume this is 18S rRNA, because no mRNA should give you a Cq of 8. If it does, you are using vastly too much cDNA.

On the other end of the scale, a Cq of 30 is...really low expression: "barely there, maybe 10 molecules per well" type expression. This will be inherently variable due to stochastic partitioning, and also very susceptible to tiny changes in transcriptional noise.

The difference between these two is 22 cycles, which corresponds to about 4,000,000-fold difference. That's...a lot. You are attempting to normalise something that is barely there to something that is probably the dominant target in your entire sample, and that's...not a great way to do this.

Plus, when one amplicon needs 22 additional cycles, efficiencies really, really start to come into play (which might explain the difference between your data).

As an example,

22 extra cycles of a reaction at efficiency 2.0 = 4.19e6 fold amplification.

22 extra cycles of a reaction at efficiency 2.16 = 2.2e7 fold amplification.

That tiny difference in reaction efficiency compounds at every step and results in a five fold difference in amplification (which in practice means your measured Cq values represent an overestimation of your target).

Using a reference that is vastly in excess of your GOI really hurts you, here, basically.

So I am going to say, at this point, that you need to go back to more basic stuff before worrying about discrepancies between two normalisation methods.

I am not convinced your reference is an appropriate choice (and I would further recommend you use at least two references).

I am not convinced your target gene is meaningfully expressed anyway.

Based on these two facts, I wouldn't trust either normalisation method, because I wouldn't trust the underlying data.

How did you determine your reference gene (and is it 18S)?

What other references did you test?

How do you prepare your RNA, and how do you QC it?

Do you DNAse treat your RNA?

How much RNA do you use for cDNA synthesis?

How much cDNA do you use per well for qPCR?

How did you design your primers?

If using SYBR green, do you run melt curve analysis?

Do you have positive controls (samples you know express your GOI)?

Do you include negative controls (both "no template" and "does not express GOI")?

Thank you, your answer is very helpfull.

RNA was isolated using a ZYMO RNA kit, than it was checked with NanoDrop™ Lite Spectrophotometer. DNAse is included in the used cDNA synthesis kit (Termo Scientific). I equalize RNA before the reverse transcription to 500 ng/µl and use 5µl and after the reaction I get cDNA in amount within 1900 ng/µl. For the qPCR reaction I prepare dilutions of 50 ng/µl cDNA and I took 1 µl for 20µl reaction mix, so 50ng cDNA was used for one reaction.

I used SYBR Green so curve analyses were done. I think that it looks good for both target and reference gene - a clear single product.

The target gene expression is growing noticeably in the 3-4 day after infection (it is use as positive control). I include negative control "no template", the negative control without the target gene was used at the beginning of my work, now I don't add it.

And you have right, my reference gene is 18S. I also chacked elongation factor-1alpha (Ef1α), elongation factor-2alpha (Ef2α) and tubulin (57-α-tubulin).

Now I plan to check glyceraldehyde-3-phosphate dehydrogenase (GAPDH), adenine phosphoribosyl transferase (APRT), 60S ribosomal protein L8 (L8), Cullin 3A (CUL3A), and exocyst complex component sec3 (sec3), maybe one of them will be beter.

Hi Dorota Milczarek : this is all very helpful.

A few tips: quantifying cDNA is not (in most circumstances) meaningful.

Remember, to make cDNA you add RNA, you add primers (lots of primers), and you add dNTPs. At the end of the reaction, all of those are still there, and all of those absorb at 260.

Consider, you are adding (by my calculations) ~2.5ug of RNA to your reaction, using 5ul of RNA. I thus assume the total reaction volume is going to be greater than 5ul, and is probably 20ul (as this seems to be standard for cDNA). If you have 1900ng/ul cDNA, that means you have somehow converted 2.5ug of RNA to 38ug of cDNA: a fifteen fold amplification, somehow.

This isn't really how cDNA synthesis works.

Instead you should assume your cDNA synthesis is approximately 1:1, so if you add 2.5ug of RNA, you get 2.5ug of cDNA.

Assuming 20ul reaction vol, this means your actual cDNA concentration is closer to 125ng/ul.

Conveniently, though, you're still ok: if you're diluting your cDNA the way it sounds, you're basically diluting it 1:20 and using 1ul, so are using about 6ng of cDNA per well.

This is an entirely acceptable amount of cDNA to use (I typically use 8ng, so you're easily within the right ballpark).

So you can probably trust your numbers.

Now, for normalisation, yes: 18S is not necessarily the best choice, both because of abundance, and also because rRNAs are not subject to the exact same degradation/synthesis pathways as mRNAs, so might respond differently to your experimental conditions.

Consider: if 85% of your sample is ribosomes, and 5-10% is tRNAs, mRNAs are only a tiny fraction (5-10%) of your sample. Huge changes in total mRNA content might be masked entirely if you use 18S, because these huge changes are still only a tiny fraction of total RNA.

I would recommend using a couple of stable mRNAs rather than 18S.

Finally, all this said and done, what you are really interested in is the difference in gene expression, and ddCt methods seem almost entirely designed to hide this from you behind a wall of inexplicable numbers and log transformations.

What I like to do (because I can factor in efficiencies this way) is convert all numbers from log values (Cq) to linear relative quantities (RQ).

For this, for each gene, you take the lowest Cq value (i.e. the sample with the highest expression of that gene), and then for each sample:

RQ = efficiency (lowest Cq-sample Cq)

In other words, your highest expressing sample will now be 1 (lowest Cq - lowest Cq equals 0, and anything to the power of 0 is 1), and everything else will be linearly related to that (so a sample with half as much will be 0.5).

This will factor in your efficiency differences, so here you would use 2.16 as efficiency, rather than 2.

You can do the same for your references. Your normalisation factor is the geometric mean of these.

Then you normalise your data by simple division, because we're now in linear space.

Your final (normalised) data should then be log transformed again for stats and plotting, because qPCR data is typically log-normally distributed, but it will now be normalised.

NOTE:

If you know your efficiencies are equal between reactions, you can short cut this all and simply generate dCt values: sample Cq - refgene Cq (use the arithmetic average of reference Cqs if using multiple references).

dCT values are mathematically identical to log transformed normalised RQ values obtained via all the faff described above, so if you can use them, it's much quicker.

For all stats, just compare these numbers. Don't generate ddCt values, because all these tell you is the fold change, which generally you can just determine by looking at the plots.

Stats will tell you whether your expression change is statistically significant.

Your own understanding of your biological system will tell you if the magnitude of this change is biologically meaningful.

Thanks John Hildyard, now I can to work on it :-)