Figure 6

The density distributions of the outlierness value, O , and the corresponding false discovery rates from the benchmark results for HIV-1 and HCV nucleotide sequences. For all the HIV-1 (a) and HCV (b) sequences used in the benchmark tests (Table 1), the O values were surveyed and plotted as the histograms that were separately normalized for the cases concordant with (bar) and discordant to (line) LANL genotypes/subtypes. After filtering out the cases having O > cutoff, the discordant ones were counted as false positives. The false positive rates and the proportion of the sequences retained (coverage) were plotted against the O cutoff for HIV-1 (c) and HCV (d) sequences. The suggested cutoff is shown by a dashed line. The revised plots after removing the HCV sequences of suspicious genotypes are available in Supplementary Figure 5 in Additional File 3.