Despite its name, "digital PCR" does not refer to PCR conducted on a computer. Rather, the term "digital" arises from the binary nature of the results obtained from the tiny sub-reactions, results that are positive or negative, forming a digital code. The term also refers to its design as a non-analogous method, whereby results are an absolute, digital value, without the need for comparison to an individual standard curve.
Generally, dPCR reactions comprise many parallel reactions that are occurring as either droplets, nano-wells, or in our case, thousands of nanoreactor Beads held in suspension and separated from each other.
An individual reaction takes place in each of these sub-volumes, dependent upon the presence or absence of target and the according primers (and if applicable additional probes). The result of these PCR reactions is either negative or positive, in the latter case regardless of whether there was one, three or 100 copies in a nanoreactor. Unlike qRT-PCR and its Ct-value determination, it does not matter by design when the dPCR reaction becomes positive but rather how many of the thousand parallel reactions turn positive. In contrast to qRT where time-consuming imaging is performed after every cycle, BLINK dPCR runs for 45 cycles with imaging performed at the very end of the PCR reaction.
Digital PCR does not need the information at which cycle a PCR product reaches a threshold concentration because as we split the sample up into thousands of sub-reactions we can apply the statistics of large numbers, namely the Poisson analysis. With its help the mean number of target copies the nanoreactors contained at the beginning of the reaction can be assessed.
The distribution of target molecules over the beads is described by the Poisson distribution. Following PCR Beads will be positive (bright) if they contain one or more target molecule or they will stay negative (dark) if they contain no target molecule. The Poisson analysis of digital PCR calculates the mean number of target copies by simply knowing the proportion of negative beads. The mean number of target copies per bead is called “lambda”.
In Figure 2 the bright Beads have seen target. This equates to approximately every fifth or sixth Bead. However, simply counting the positive beads does not give the correct target number, because beads can have seen more than one target molecule. For this reason, the Poisson distribution is applied which automatically takes the possible occurrence of multiple target molecules in any individual bead into account.
.jpg)
Figure 3 demonstrates two peaks - a darker negative fraction and a lighter positive fraction; this is used for estimating a threshold which can also be corrected manually. All Beads are checked against validity criteria (e.g. bead width, brightness distribution); if they meet the criteria they are used for the analysis. Beads which are neither clearly positive nor negative are termed “rain” as they appear to drip from the upper positive to the negative fraction.
Beads with signal intensity below the threshold are negative. The proportion of negative beads to the overall number of beads is applied on the Poisson distribution, to achieve the lambda.
Figure 4 shows a dilution series over 5 orders of magnitude, clearly demonstrating an increase in the number of positive beads as the concentration of target increases. Additionally, the number of mean target molecules per Bead is also rising, resulting in several target molecules per positive Bead. After the PCR reaction, positive Beads turn bright due to the unquenched probes or intercalators independent of the starting template concentration as the reaction is brought to saturation with 45 cycles. This illustrates the underlying mechanism of dPCR analysis.
.jpg)
The measuring range of the digital PCR largely depends on the number of beads. An assay with 20,000 beads can reach a dynamic range of five orders of magnitude.
The analysis of the sheer quantity of thousands of parallel reactions with the help of the Poisson distribution allows highly reproducible results over several log levels. This explains why digital PCR is not only valued for typical nucleic acid test but is especially advantageous in assessing rare events (like rare mutation detection)or slight differences (such as copy number variation or transcript profiling).