Sensitivity and specificity comparison of ViVaMBC with pileup of a clinical HCV sample.
a) Comparison of the codon frequencies after piling up the data (x-axis) with the estimated frequencies of ViVaMBC (y-axis). Codons represented with triangles were absent after 454 sequencing on the same sample and hence assumed to be false-positive findings. Codons colored in grey are present in either one of the two methods. Frequencies of 0.5% and 0.25% are indicated with dotted and dashed lines respectively. Above 0.5% and even above 0.25% a good correlation is observed where a few false-positive findings are filtered out using ViVaMBC b) False discovery rates for both ViVaMBC and pileup are calculated with changing reporting limits. The FDR is higher and increases more rapidly for the pileup.