TABLE 3.

Comparison between haplotype and genotype data: performance for models with uniform recombination rates with unphased data

γa	ρa	g(γ)b	g(ρ)b	B(γ)c	B(ρ)c	E()d	E()d	V(γ)e	V(ρ)e
Hudson's composite-likelihood method
20	20	0.452	0.968	0.600	0.573	22.060	20.170	1.273	0.495
60	20	0.692	0.954	0.545	0.491	55.259	21.594	0.650	0.573
80	20	0.794	0.962	0.626	0.543	74.119	20.908	0.534	0.564
20	40	0.420	0.968	0.662	0.484	19.919	42.000	1.345	0.426
40	40	0.572	0.960	0.619	0.534	36.620	40.880	0.863	0.417
Summary statistics method
20	20	0.580	0.910	0.445	0.498	27.600	22.700	1.387	0.715
60	20	0.732	0.850	0.454	0.453	70.278	24.090	0.763	0.822
80	20	0.748	0.838	0.468	0.514	89.940	23.226	0.654	0.872
20	40	0.488	0.904	0.516	0.521	28.076	42.582	1.673	0.512
40	40	0.600	0.828	0.501	0.541	51.814	43.050	1.169	0.590

^aγ and ρ denote the true values of the gene-conversion and crossing-over rates under which 500 data sets were simulated.

^bg(γ) and g(ρ) denote the fraction of the data sets for which the estimates of gene conversion () and crossing over () lie within a factor of 2 of the true values (i.e., γ and ρ), respectively.

^cB(γ) and B(ρ) denote the fraction of times the estimates of gene conversion and crossing over are lower than the true values, given that they are not equal to the true values.

^dE() and E() denote the mean of the estimates of gene-conversion and crossing-over rates.

^eV(γ) and V(ρ) denote the root mean square relative error for the estimates.