Figure 2.—
Likelihood curves obtained by SweepFinder (solid lines) and predicted hidden sequences obtained by HMMB (dashed lines) and HMMB-SEG (dashed-dotted lines) for two samples of size n = 30 and length L = 100 kb generated with α = 300 and τ = 0.001. For SweepFinder, the 95% threshold was 10.00 in this case. The states “neutral,” “intermediate,” and “sweep” are represented, respectively, by the values 4, 8, and 12 for HMMB and by the values 10, 14, and 18 for HMMB-SEG. The true selected site is in the middle of the segment, at position 50 kb. Large differences between SweepFinder and HMMB, as for sample 2, were observed in ∼4% of the simulated samples.