ISHAPE: new rapid and accurate software for haplotyping

BMC Bioinformatics

Table 1 capture rate and number of haplotypes detected by the various algorithms

1A :
		0% MD		2% MD		5% MD		10% MD
Ishape	APOE	1.00	32.7	1.00	43.6	1.00	53.4	0.99	70.4
Phase2.1		1.00	29.0	1.00	32.5	0.99	35.4	0.99	40.1
Phase1.0		0.99	29.0	0.99	27.8	0.99	29.6	0.98	32.5
FastPhase		0.89	29.0	0.86	27.2	0.82	27.2	0.77	27.7
PL-EM		0.89	20.0	0.90	21.1	0.89	21.3	0.88	21.1
Ishape	GH1	0.99	101.5	0.99	148.2	0.98	229.6	0.97	365
Phase2.1		0.98	71.0	0.97	70.5	0.97	74.3	0.96	83.2
Phase1.0		0.97	46.0	0.96	50.2	0.95	54.1	0.93	59.7
FastPhase		0.88	55.0	0.87	52.4	0.82	53.1	0.76	54.2
PL-EM		0.91	41.0	0.90	42.4	0.89	43.4	0.86	42.5
1B :
	APOE (9 SNPs)				GH1 (14 SNPs)
	0%	2%	5%	10%	0%	2%	5%	10%
Average number of possible configurations per genotype	3.25	4.48	8.21	22.89	9.62	18.69	48.72	244.13
Average number of candidate configurations per genotype + (ICR)	1.59 (1.0)	2.3 (1.0)	3.0 (1.0)	4.8 (0.99)	2.3 (0.99)	3.31 (0.99)	5.4 (0.98)	10.2 (0.97)

A) Comparison of the algorithms'performance on the APOE and GH1 datasets regarding the average ICR (left column) and the average number of detected haplotypes (right column). For each level of missing data (0%, 2%, 5%, 10%), 100 experiments were performed. Best performances are highlighted in bold.
B) Estimation of the reduction of the space of possible haplotypes for the GH1 and APOE genes operated thanks to the Bootstrap-IEM. The values are the average of 100 experiments. In the lower line, the values in brackets give the correspondingaverage ICR.

ISSN: 1471-2105