As COVID patients can vary greatly in their viral load, it is important to have a primer set that can amplify extremely low copies (<10) of template per reaction. This is especially troublesome when you are using long primers that contain indices and Illumina adaptor sequences in conjunction with the primer sequence.
After some initial primer optimization (detailed elsewhere), we identified two promising primer sets and defined their LoD by RT-qPCR.
In this experiment, we tested a gradient of primer concentrations of two primer pairs with full sequencing adaptors against a gradient of copies of synthetic SARS-CoV-2 RNA from Twist in a background of human cell line lysate. The primer pairs are from the CDC assay (N1) and from this article (S2). Both primer sets had amplification distinguishable from the no template down to 10 copies per reaction at 400 nM.
Full results are in the experiment page, but gels looked promising. However, for the N1 set no template had a similar size to the amplicon, whereas S2 looked distinct.
We next looked at our limit of detection with the S2 and N1 primer set by sequencing using a 3-fold dilution of purified SARS-CoV-2 RNA from ATCC spiked in at known molecule counts in a background of human cell line lysate (where we are detecting RPP3 in parallel) and a synthetic RNA spike-in with the same priming sequences as the SARS-CoV-2 RNA. The y-axis is the ratio of purified genomic RNA to the synthetic spike-in (100 copies/reaction). We set an empirically determined significance threshold based off our no RNA controls. This enabled us to detect down to ~5 molecules/reaction for the S2 set and ~15 molecules/reaction for the N1 set. The N1 set is likely limited by our characterized contamination of primer sequences described here:
Detection calls are have p = 0.019 and p=0.022 for the N1 and S2 primer sets, respectively.
Detection calls are have p = 0.019 and p=0.022 for the N1 and S2 primer sets, respectively.