The baseline is acceptable, the threshold is adjusted properly, the NTCs are negative, the positive control is good - now you are ready to look at the samples. First, look at the amplification curves. What is the ARn at the plateau? If the ARn is not greater than 0.5, then there may not be real amplification. The amplitude of the ARn depends upon several factors. One important factor is the probe. Each probe may perform differently and have different maximum ARn values. The type of probe and what it is labeled with can affect the ARn substantially as can the type of master mix that is being used. Figure 2.10A, shows an example of the two different ARn plateaus that are obtained using a 5' FAM, 3' BHQ1 probe (Biosearch Technologies, Novato, CA, USA) and a MGB probe from a commercial Gene Expression assay (Applied Biosystems, Foster City, CA, USA). Checking the Raw Spectra view after 40 cycles of amplification (Figure 2.10B) reveals that there is a substantial difference in the raw fluorescence generated by each probe. Further investigation into the Multicomponent view (Figure 2.10C and 2.10D) demonstrates that even though the raw fluorescence values are quite different, there is exponential amplification using both probes. As long as the threshold is set so that it is in the exponential phase of both sets of curves, analysis should be valid.
Look at the samples' curves again. Are all the curves exponential or are there some curves that run parallel to the threshold before becoming exponential? This can happen when the ROX is gradually dropping, as discussed in the NTC section and demonstrated in Figure 2.9A. Curve b. It is also possible to have a run with no real amplification that looks like there are amplification curves. The software will do its best to make curves. However, if the ARn is much less than 1, then it should immediately be suspected that no real amplification has taken place even though it looks like curves have been generated. Figure 2.9A. Curve c, which has already been described in the NTC section, demonstrates this situation. If a good positive control was included, then it should be easy to distinguish real amplification from machine artifacts.
Check the replicates next. They should be within 0.5 Ct of each other. Determine the coefficient of variance. What is the coefficient of variance (CV)? The CV is the standard deviation (SD) divided by the arithmetic mean and is used to measure intra-assay reproducibility from well to well and is also useful to measure inter-assay variation from assay to assay. If the CV for the samples is small, then there is no problem. What if one well is different from the others? Hopefully, triplicates were run. Here again, the Multicomponent view or Raw Spectra view is useful. If there was no amplification, check for the presence of the probe fluorescence. Maybe the probe did not get added? If there is no sign of probe fluorescence, then this would be a legitimate reason to exclude this data. If the probe is present, but there is no sign of amplification, perhaps no template was added. Similarly, if the aberrant well has a lower Ct, then perhaps template was added twice. Of course, adding template to the same well twice would only decrease the Ct by one, so if the CV is large, this phenomena would not be detected. Another scenario is that the plate was not sealed tightly and there has been evaporation in one or more of the wells/tubes. The Multicomponent view might show all fluorescent compounds increasing over the course of the run due to evaporation. Using genomic DNA as a template can be especially problematic for getting identical replicates and careful attention needs to be paid to mixing between every aliquot. If dealing with samples that have low expression and therefore a very high Ct, then the Monte Carlo effect should be considered. Try to run triplicates and run the same experiment three times if possible, to gain statistical significance for the results (Bustin and Nolan, 2004b). As a general rule, the more replicate reactions there are, and the lower the CV of the replicates, the better the ability to discriminate between samples.
Was this article helpful?