Out of the 23,250 bases sequenced in the calibration amplicons, we detected an average of 183 significant variants, indicating the assay specificity is greater than 99.9% (Table S5 in Additional file 2). Across all calibration-tested sample combinations, the average sensitivity was 89.1% (± 3.3%) when significant variants at 1% or greater prevalence are considered (Figure 1a). As anticipated, the sensitivity is better (> 94%) for mutations present at 5% or greater than for mutations present at 1% prevalence (75%) (Figure 1b). The average positive predictive value (PPV = 1 - (False positive rate)) is 97.6% (± 1.9%) with a noticeable lower PPV (90.5%) when expected prevalence is less than 5% (Figure 1c). As illustrated in Figure 1d, the observed prevalence is highly correlated (Pearson correlation r = 0.97) with the expected one; thus, the prevalence of the mutant allele in the DNA sample can be accurately estimated.
