Hi,

first things to do is normalize all your samples including your control to the house keeping gene. then, you normalize your treatment to the control. stranded deviation is so important as it show you the efficiency of your triplicate also the consistency of your experiment, so it is so necessary.

The data you show is for a single dCt value (even if this is determined as an average from several technical replicates) from a single biological sample (actually two: one "control" and one "treatment" [assuming the last line is a typo and should read "Ref treatment"]).

If this is all you have, then there is no way to estimate the SD between samples. The SD between technical replicates is usually not interesting (exception: if you want to assess the quality/reliability/reproducibility of an assay - what does not seem to be your question).

If you have a group of more than one biological sample, you can calculate the SD of the dCt values for these samples as usual.

Assuming that the these values are the Ct values, and that "Ref control" is "Ref treatment" as Jochen Wilhelm suggested, I don't see why we can't calculate a SD! For example, like I did here (https://gist.github.com/MahShaaban/52753575dd951678c7b5425e437f9ecb)

Note: the range of the variance in the reference gene Ct value is very big and would probably cause a big error term.

I'd be very interested to see what Jochen Wilhelm thinks about the example above and whether Eline ten Dolle thinks this answers the question.

Although it is technically possible to calculate the SD of dCt values for a single sample, it is not required for the analysis and it is not useful to interpret the results. To control the homogeneity of technical replicates, the variability of Ct values (not dCt values) should be analyzed, and with only n=3, the SD is not quite a robust measure (however, it still may point out samples with obvious problems). It's simpler and more robust to check the ranges of Ct values instead.

What is interesting is the uncertainty associated with ddCt value. Roughly: in what range should I expect the ddCt value if I would repeat the experiment with independent samples? The technical variation seen in that particular sample that has been analyzed does not contain any information about how another, independent sample would behave. Therefore it's rather useless for the interpretation of these results what the SD of the tech. reps. is (when I am right that the data is only about a single sample [1 control + 1 treated]).

Thanks Jochen Wilhelm

If I understand correctly, you describe a situation where the three measurements are of the same samples (technical replicates). Right? It is not clear to me from the original post whether the triplicates are biological or technical. If it's the case that the measurements are from three independent samples, does the calculation hold?

Yes, you understand correctly. My answer was based on the assumption that the three values (for goi and ref within each group) were from technical replicates. The reason why I made that assumption is that Ct values were given and not dCt values.

And yes, that's not clear from the OP. It was my assumption.

If the three values are from three independent samples/individuals, then it is not stated what values for goi and ref are from the same individual. I would have to assume (again), that the order in all cases is the same (1st -> 2nd -> 3rd individual) to calculate meaningful dCt values. Under that assumption, the SD for the control dCt values is 0.31, and for the treated samples it is 1.85. The pooled SD is 1.33.

Alight. Thanks again, Jochen Wilhelm