qPCR for CNV validation

[shiny]

I use qPCR for validation of CNVs detected by array technology. Pfaffl method is used for calculations (delta-delta Ct method corrected for PCR effeciency). In theory in case of deletion the ratio target/reference must be 1/2=0.5, in case of duplication 3/2=1.5 and in case of normal copy number 2/2=1. But in real life the tnings are a little bit more complex and I don't get ratios 0.5, 1 or 1.5. What are these boarders when I still can say that it is a deletion/duplication/normal?
It is clear that in my case the set up of specific assay for each CNV and its confirmation using multiple samples is not suitable (because the analysed CNV may be unique and present in only one sample). It is not possible to calculate the SDs.
I haven't found this kind of information in literature. Most researchers use qPCR for gene expression or DNA copy number estimation in specific regions.

[JackBean]

I think that should be similar to calculating the gene copy number etc., shouldn't it?

How much can it differ? Is it like +-0.1 (i.e. 0.4-0.6; 0.9-1.1 and 1.4-1.6) or is it like any number?
Can be your sample unhomogenous? Like some cells having more and some cells having less chromosomes?

[shiny]

Is it really +/- 0.1? At the moment I really use 0.4-0.6; 0.9-1.1 and 1.4-1.6 for deletion, normal and duplication. But I'd like to be pretty sure that these borders are widely used and approved. Could you give me some references confirming this?
About sample homogenity: they are homogenous not mosaic ones.

[JackBean]

no, I was asking, what are your numbers
The point is, that you can have some variations and errors, either due to sample problems or some working problems So, if the variation from 0.5/1/1.5 is just small, that should be fine, but if you can get like 1.3, than you can hardly say, whether is it 1 or 1.5

[shiny]

And my question was how can I calculate this errors If I only have one single sample with a specific deletion/duplication. It is not possible to calculate standard deviations I suppose..

[canalon]

You could calculate standard deviation for the essay. Do a triplicate* (or more) of your tests and of the standard and see what kind of variation is observed. That would at least give you an idea if the .1 variation is within the range of error expected or strangely unlikely considering the high repeatability of your results.

* the more independent those three repeats can be the better. Within reason of course. Can you make 3 independent DNA extraction, or at least repeat the qPCR 3 times rather than just extarcting one DNA sample and aliquoting a master mix in 3 tubes.