Figure 2.—

Comparison of maximum-likelihood estimates of ρ on data simulated with different values of f (all data simulated with ρ = θ = 1/kb) using our model (a) as compared to the LS model without bias correction (b). Our model gives good estimates of the crossover rate regardless of the amount of gene conversion present. It shows little bias, and when gene conversion is present (at least in simulated data), estimates of crossover rates can be inflated when gene conversion is not taken into account.