Quantitative amplicon sequencing—An Overview
Given the lack of sensitivity of the attempted fluorescence-based
method, we developed a sequenced-based amplicon assay. We used two
methods to quantify relative input DNA from the four Clarkia .
First, we used a common approach to quantify relative abundance of input
DNA, which simply uses the relative read abundance following the full
PCR. We refer to the traditional sequencing approach—using the
relative read abundance of amplicons at-or-near the PCR plateau
phase—as “RRA-plateau” or “RRA”. Second, we used an approach that
was specifically designed as an attempt to control for some of the
biases introduced by PCR. We contrast the RRA method with our method
that utilizes a PCR cycle treatment, which we refer to as “quantitative
amplicon sequencing” (qAMPseq).
The premise of qAMPseq applies the theory of quantitative PCR (qPCR,
A.K.A. real-time PCR) with the ability to individually index, multiplex,
and sequence hundreds of metabarcoded samples (Figure 2). Quantitative
PCR analysis uses a pre-determined threshold when the PCR reaction is in
an exponential phase of amplification, because the PCR cycle where a
reaction product moves into the exponential phase is directly related to
the starting DNA concentration, unlike the plateau stage (Kubista,
2005). Realtime PCR uses fluorescence (e.g. TaqMan chemistry) quantified
throughout thermocycling to determine the ‘cycle number’ where the
product fluorescence is higher than a background level, as the product
is in the exponential amplification phase. The estimated number of PCR
cycles when the product hits this threshold is known as threshold cycle
(Ct). This Ct value can then be compared across samples to compare
starting DNA concentrations.
In qAMPseq, we generate the same PCR amplicon in quadruplicate, with the
same starting conditions, but across different PCR cycling numbers (e.g.
20, 25, 30, and 35 cycles; Figure 2B). Subsequent cleanup and indexing
steps preserve the relative DNA amounts in each of these reactions,
which are then individually indexed (i.e. each original sample has four
unique indexes, which correspond to the different cycle ‘treatments’)
and then pooled and sequenced with all other samples (Figure 2C).
Samples can then be de-multiplexed (Figure 2D) and, within each sample
and treatment, reads are assigned to predicted taxonomic units
(“OTUs”; in this case, the four Clarkia species). The read
abundance across each sample and OTU can then be used to calculate Ct
(Figure 2E), and a more robust value the relative contribution of input
DNA (Figure 2F).