Fig. 5

Precision and recall of our method derived from numerical simulations. The x-axis shows the proportion of genomes covered by primary orthologous residue: genome bases that are occupied by conserved primary orthologs, indicating a measure of the evolutionary distance between the diverged lineages. Vertical dashed lines indicate the evolutionary distances estimated for the human-chimpanzee (~ 93%) and human-mouse (~ 35%) genomes. Different subfigures show the simulations under different mutation-duplication ratio μ/ν: (a)-(b) μ/ν = 0.001; (c)-(d) μ/ν = 0.01; (e)-(f) μ/ν = 0.1; and (g)-(h) μ/ν = 1. In different subfigures, we use the same symbols to represent the same datasets: solid squares for Lastz net alignment, crosses for BLAST RBHs, hollow circles for non-nested exact matches (“non-nested EMs”), and solid circles for “non-nested EDMs”. For the latter, the minimal length of a single exact match within each EDM is set to 20 bases for all subfigures, and the maximum length of a single run of contiguous mismatches is set to d nucleotide bases: d = 2 for subfigures (a)-(b), d = 5 for (c)-(f), and d = 15 for (g)-(h). Curves in the figure are obtained by averaging over 10 realizations; the same curves with standard deviations indicated by error bars are shown in Additional file 1: Figures S3, S4 and S5