Table 2.

. Empirical results for different strategies using the haplotype frequencies from the CAMP data with 682 trios

Strategy	β	FBAT-C Linear			FBAT-C Robust			Bonferroni
		exactly all DSLs	at least one DSL	only non-DSL markers	exactly all DSLs	at least one DSL	only non-DSL markers	exactly all DSLs	at least one DSL	only non-DSL markers
1 DSLa	0.25	0.396	0.032	0.150	0.392	0.112	0.150	0.238	0.028	0.040
	0.5	0.832	0.082	0.084	0.814	0.194	0.104	0.118	0.078	0.000
2 DSLa	0.25	0.242	0.470	0.052	0.116	0.540	0.084	0.026	0.646	0.000
	0.5	0.672	0.300	0.000	0.450	0.362	0.028	0.088	0.794	0.004
Untyped DSLb	0.25	0.372	0.058	0.064	0.354	0.056	0.066	0.352	0.166	0.046
	0.5	0.834	0.162	0.002	0.846	0.132	0.016	0.254	0.744	0.002

When the true model is one DSL, the column ‘at least one DSL’ indicates the case when the DSL is found and additional non-DSL markers are found. When the true model is two DSLs the column ‘at least one DSL’ included the case when only one of the two DSL markers is found, with or without any additional markers. The columns for each strategy do not sum to 1, as the case when no markers are found is not included in the table.

^aFor 1 and 2 DSLs, random markers were chosen as the DSL(s).

^bFor the untyped DSL, rs1800872 was chosen as the DSL, but was omitted from the marker set, and rs1800871 was considered to be ‘the true DSL’ because of the high correlation (0.93). Based on 500 simulations; approximate SE <0.01.