. 2003 Oct 20;73(5):1162–1169. doi: 10.1086/379378

Table 2.

Error Rate and Switch Error Rate for the Data Sets Considered by Lin et al. (2002) ^[Note]

	Error Rate for Data Set
Algorithm	GLRA2	MAOA	KCND1	ATR	GLA	TRPC5	BRS3	MECP2
Lin et al. (2002)	.79	.61	.54	.62	.89	*.58*	.72	.85
HAPLOTYPER	.89	.76	.72	.72	.79	.72	.79	*.64*
Modified PHASE	*.76*	*.54*	*.46*	*.45*	*.68*	*.58*	*.67*	.77
	Switch Error Rate for Data Set
Lin et al. (2002)	.14	.10	.22	.29	.22	*.13*	.14	.23
HAPLOTYPER	.16	.12	.27	.32	.16	.20	.15	.19
Modified PHASE	*.10*	*.07*	*.13*	*.18*	*.11*	*.13*	*.10*	*.15*

Note.— The results for HAPLOTYPER and the Lin et al. (2002) algorithm are taken from table 1 in Lin et al. (2002). The results for PHASE were obtained by us on 100 data sets simulated in the same way as those used to produce table 1 in Lin et al. (2002) (i.e., by randomly pairing the 40 X-chromosome haplotypes used by Lin et al. [2002], kindly provided by D. Cutler). Each number in the table is based on results for 100 data sets. For example, the error rates are the total number of mistakes made across all 100 data sets, divided by the total number of ambiguous individuals in all 100 data sets. The results for the best-performing method in each column are in boldface/italics.