Skip to main content

Table 2 performance of the various algorithms on the GH1 and APOE datasets

From: ISHAPE: new rapid and accurate software for haplotyping

2A. GH1 dataset

Soft

MD

IF

IER

Time (sec.)

MD

IF

IER

Time (sec.)

Ishape1

0%

0.927 +/- 0.001

0.119 +/- 0.001

0.9

5%

0.915 +/- 0.002

0.164 +/- 0.004

1.7

Ishape2

 

0.937 +/- 0.001

0.103 +/- 0.001

9.2

 

0.927 +/- 0.002

0.147 +/- 0.004

11.5

Phase2.1

 

0.937 +/- 0.001

0.103 +/- 0.001

62.9

 

0.924 +/- 0.002

0.148 +/- 0.004

71.4

Phase1.0

 

0.926 +/- 0.002

0.119 +/- 0.002

15.6

 

0.915 +/- 0.003

0.164 +/- 0.005

26.0

FastPhase

 

0.928 +/- 0.001

0.105 +/- 0.001

139.1

 

0.920 +/- 0.002

0.170 +/- 0.004

138.9

PL-EM

 

0.915 +/- 0.001

0.116 +/- 0.000

0.3

 

0.890 +/- 0.003

0.171 +/- 0.003

3.2

2snp

 

NA

0.157 +/- 0.000

< 0.1

 

NA

0.214 +/- 0.002

< 0.1

Ishape1

2%

0.922 +/- 0.001

0.137 +/- 0.003

1.2

10%

0.905 +/- 0.002

0.208 +/- 0.004

2.8

Ishape2

 

0.933 +/- 0.001

0.120 +/- 0.002

10.6

 

0.916 +/- 0.002

0.195 +/- 0.005

14.6

Phase2.1

 

0.931 +/- 0.001

0.122 +/- 0.003

64.6

 

0.914 +/- 0.002

0.196 +/- 0.005

82.5

Phase1.0

 

0.921 +/- 0.002

0.138 +/- 0.003

20.7

 

0.903 +/- 0.003

0.211 +/- 0.005

33.9

fastPhase

 

0.924 +/- 0.001

0.134 +/- 0.004

147.5

 

0.907 +/- 0.002

0.241 +/- 0.006

134.6

PL-EM

 

0.913 +/- 0.003

0.140 +/- 0.003

1.0

 

0.854 +/- 0.004

0.225 +/- 0.005

12.6

2snp

 

NA

0.176 +/- 0.002

< 0.1

 

NA

0.283 +/- 0.004

< 0.1

2B. APOE dataset

Soft

MD

IF

IER

Time (sec.)

MD

IF

IER

Time (sec.)

Ishape1

0%

0.946 +/- 0.001

0.062 +/- 0.001

0.2

5%

0.932 +/- 0.003

0.109 +/- 0.005

0.4

Ishape2

 

0.941 +/- 0.001

0.057 +/- 0.001

3.5

 

0.926 +/- 0.003

0.102 +/- 0.005

4.1

Phase2.1

 

0.940 +/- 0.001

0.055 +/- 0.001

14.0

 

0.923 +/- 0.003

0.102 +/- 0.005

15.8

Phase1.0

 

0.947 +/- 0.001

0.062 +/- 0.000

2.7

 

0.932 +/- 0.003

0.108 +/- 0.005

3.9

fastPhase

 

0.876 +/- 0.001

0.118 +/- 0.002

49.1

 

0.870 +/- 0.003

0.181 +/- 0.005

44.2

PL-EM

 

0.897 +/- 0.000

0.125 +/- 0.000

0.1

 

0.883 +/- 0.004

0.159 +/- 0.005

0.4

2snp

 

NA

0.200 +/- 0.000

< 0.1

 

NA

0.227 +/- 0.004

< 0.1

Ishape1

2%

0.942 +/- 0.002

0.078 +/- 0.003

0.3

10%

0.917 +/- 0.004

0.149 +/- 0.007

0.6

Ishape2

 

0.935 +/- 0.002

0.070 +/- 0.003

3.9

 

0.910 +/- 0.004

0.143 +/- 0.007

4.6

Phase2.1

 

0.933 +/- 0.002

0.072 +/- 0.003

14.8

 

0.907 +/- 0.004

0.146 +/- 0.007

17.4

Phase1.0

 

0.941 +/- 0.002

0.078 +/- 0.003

3.2

 

0.917 +/- 0.004

0.150 +/- 0.007

5.1

fastPhase

 

0.875 +/- 0.002

0.140 +/- 0.003

47.0

 

0.864 +/- 0.004

0.225 +/- 0.007

45.4

PL-EM

 

0.894 +/- 0.003

0.137 +/- 0.003

0.2

 

0.854 +/- 0.005

0.191 +/- 0.006

1.3

2snp

 

NA

0.208 +/- 0.002

< 0.1

 

NA

0.259 +/- 0.004

< 0.1

  1. Different missing data levels are tested, each with 100 experiments. The mean accuracy (IF and SER) and runtime of the haplotyping algorithms are compared on A. the GH1 dataset, B. the APOE dataset. The 95% confidence intervals are also given. Best performances are highlighted in bold. For 2-SNP, the software does not provide haplotype frequency estimation: thus, the IF is not available (NA).