Skip to main content

Table 5 Concordance between genotype calls with different levels of conservativeness from low- and high-coverage sequence data, using allele read counts obtained from aligned reads in BAM files

From: Impact of index hopping and bias towards the reference allele on accuracy of genotype calls from low-coverage sequencing

 

na

Genotype concordance (%)

Allele concordance (%)

Concordance by genotype (%)

True = 0

True = 1

True = 2

0|0b

1|0

2|0

0|1

1|1

2|1

0|2

1|2

2|2

Best-guess

 1×

30,875

62.6

81.1

99.45

0.55

51.35

48.65

0.70

99.30

 2×

35,688

79.9

90.0

98.60

1.40

0.00

26.20

48.26

25.54

0.04

1.61

98.35

 3×

28,357

88.6

94.3

98.36

1.64

0.00

13.80

72.20

14.00

0.00

2.22

97.78

 4×

17,849

92.7

96.4

98.05

1.95

0.00

8.22

84.01

7.77

0.00

2.63

97.37

 5×

9619

95.3

97.6

97.40

2.60

0.00

4.55

91.52

3.93

0.00

2.49

97.51

 6–10×

7047

95.2

97.6

97.73

2.27

0.00

4.96

91.05

4.00

0.00

2.68

97.32

 Overall

129,435

81.4

90.7

98.53

1.34

0.13

24.27

52.40

23.33

0.18

1.61

98.21

Probability ≥ 0.90

 1×

0

 2–3×b

14,572

95.5

97.7

100.00

100.00

100.00

 4×

14,359

92.7

96.3

99.97

0.03

0.00

16.24

68.39

15.37

0.00

0.05

99.95

 5×

8315

96.3

98.2

99.92

0.08

0.00

6.76

87.41

5.83

0.00

0.10

99.90

 6–10×

6397

98.1

99.0

99.83

0.17

0.00

3.08

94.65

2.27

0.00

0.29

99.71

 Overall

43,643

95.1

97.5

97.18

2.82

0.00

3.52

93.26

3.21

0.00

3.65

96.35

Probability ≥ 0.98

 1–3×

0

 4–5×b

4366

99.8

99.9

100.00

100.00

100.00

 6–10×

6313

98.1

99.1

99.83

0.17

0.00

3.15

94.58

2.26

0.00

0.29

99.71

 Overall

10,679

98.8

99.4

99.65

0.35

0.00

1.00

98.28

0.72

0.00

0.57

99.43

  1. aNumber of genotypes called for 5,531 SNPs across 26 individuals
  2. bHeterozygotes are easier to call than homozygotes; at these coverages, certainty is not sufficient to call the homozygotes, but note, that the actual counts for (1|0) and (1|2) are very low compared to (1|1): 19-fold and 569-fold lower for genotypes called with a probability greater than 0.90 and 0.98, respectively
  3. Concordance is shown by coverage at variant site