Assay efficiency, specificity, and sensitivity assessment (in
vitro)
This step was facilitated by the use of a High Throughput qPCR (HT-qPCR)
platform (Biomark HD™, Standard BioTools Inc, USA), which enabled the
time- and cost-efficient in vitro assessment of a relatively
large number of makers (e.g. compared to conventional qPCR instruments).
For this step, template DNA (i.e. extracted from tissue of available
specimens) was normalized to 1.5 ng/µl per sample following
spectrophotometric quantification with a dsDNA BR (Broad Range) assay
kit using a Qubit 3.0 fluorometer (Invitrogen). In addition, to further
test specificity of Mytilus and Ensis species complexes,
synthetic double stranded DNA (GBLOCKS, Integrated DNA Technologies) of
the relevant target genes was obtained for Mytilus
galloprovincialis, Mytilus edulis, Mytilus trossulus, Ensis ensis, Ensis
siliqua, Ensis leei ( synonym Ensis directus), Ensis magnus
(synonym Ensis arcuatus) and standardized to 10,000 copies/µl.
Additionally, to act as a universal positive control and generate
standard curves for all assays
tested, a six points 10-fold
dilution series (ranging from
1,000,000 to 10 copies per µl) was created with an oligonucleotide pool
containing synthetic double-stranded DNA fragments of all target
genes/species combinations (OLIGOPOOLS, Integrated DNA
Technologies)( see supplementary material Table S1) .
The HT-qPCR runs were performed using IFC controller MX and HX (for
priming and loading Integrated Fluidic Controls, IFCs) and a BIOMARK HD™
real-time PCR system (Standard Bio Tools Inc, USA),
following manufacturer’s
recommendations, including a pre-amplification step (as per protocol QR
100-5876C2) and the actual qPCR runs using the 48.48 IFC (as per
protocol QR 100-9791Rev3) or 96.96 IFC configurations (as per protocol
QR 100-9792Rev3)(all protocols available at
https://www.standardbio.com/support/instrument-support/biomark-ep1-support).
Exceptions included the addition of of 1µg/µl BSA (Thermofisher) in the
pre-amplification reaction, and subsequent PCR reaction clean up by
addition of exonuclease at 4 U/reaction (New England Biolabs) and a
thermocycling profile including 37°C for 30 minutes and 80°C for 15
minutes. Cleaned up products were then diluted 10-fold with TE
suspension buffer (Qiagen) prior to performing the actual qPCR run on
the relevant IFC controller and Biomark HD. Final thermal cycling
followed manufacturer’s protocols and included a melt curve analysis
(i.e. “GE Fast 48x48 PCR+Melt v2.pcl” or “GE Fast 96x96 PCR+Melt
v2.pcl”). Each configuration enables the simultaneous real-time qPCR
amplification (in parallel) of either 2304 (48 samples tested against 48
assays) or 9216 (96 samples tested against 96 assays) reactions. This
allowed the screening of all target species’ template DNA in duplicate,
the standard dilutions points in six technical replicates for each
dilution point per run, including all assays (at least in duplicates) as
well as 3 Internal Amplification Controls (IACs)
(see supplementary materialAppendix S1 for a more detailed overview of the protocols
used). Following each run, qPCR data was inspected using the Real-Time
PCR analysis software version 4.5.2 (Standard BioTools Inc, USA). Output
data was exported and analysed in R-studio (Posit team, version
2022.12.0+353).
To assess sensitivity of each assay, data generated by the six points
10-fold dilution series was used to estimate Limit Of Detection (LOD)
and Limit Of Quantification (LOQ) for each assay following a discrete
threshold as well as a curve-fitting modelling approach, as described in
Klymus et al. (2020). The discrete threshold approach for LOD identifies
the lowest concentration with 95% positive replicates, while for LOQ
the lowest standard concentration that could be quantified with a CV
value below 0.35 was selected. These CVs were modeled for exponential
decay, linear, and polynomial models to select the lowest residual
standard error. When possible, a threshold of CV 0.35 was used on the
model to estimate the number of copies for modeled LOQ. Assay efficiency
was checked according to expected theoretical values (i.e. optimal
efficiency range 90-100% and a regression line slope between −3.6 and
−3.3), whereas assay specificity was assessed by checking whether
positive amplification (i.e. beyond the LOD) was observed in the target
species and/or any off-target species, as well as by melt curve peaks
comparison with a difference of >1 degree Celsius selected
as the threshold for melt-curve specificity.
After calculating the efficiency, LOD and LOQ, two methods were used to
quantify the detected DNA Cq-values to copy numbers. The first method
which is the more conventional method, is based on the obtained slope
and intercept values from the standard curve. A second method was used
to recalibrate intercepts based on the efficiency values of all the
assays (see supplementary material Appendix S2 for a more
detailed explanation). Assuming that efficiency was accurately
calculated over 5 similar concentrations for all markers in the HT-qPCR,
for the second quantification method, the intercepts were plotted
against the efficiency of the assays, which resulted in a significant
linear model. To improve the fit of the linear equation, outliers were
defined based on the standardised residuals of the linear model. More
specifically, any observation with a standardised residual equal to or
greater than 2 (in absolute value) was deemed to be an outlier and was
removed to improve the fit. The linear model was then used to calculate
an additional intercept that was used to convert Cq values into copy
numbers. The assays specificity was assessed by verifying the positive
reactions, grouped per sample/assay, using the melt curve and AMP
values. Results were displayed by creating a heatmap showing average
amplification strength in copies per/µl, with the corresponding AMP
values of the same PCR reaction. Matrices for the heatmaps were created
in R-studio with visualization done in Excel (Microsoft, Version 2208).
2.3. PHASE 3 –FIELD TESTING AND
VALIDATION