Skip to main content

Table 2 Accuracy in high-GC rich regions. Errors, Sensitivity, and FDR for the ACE, WGS, SSCR, SS, NX, and NG platforms based on evaluation of observed variant calls using data normalized to 100× mean coverage (conventional WES and ACE) or 31× WGS. Calculations are based on position and genotype matching to the GIBv2.18 less restrictive call-set within the MIG (left), a target region common to all ACE and WES platforms (middle, Common Target File), and a target region aggregated across all ACE and WES specific target files that contain moderate-impact and high-impact loci (right, Union Target File)

From: Achieving high-sensitivity for clinical applications using augmented exome sequencing

  

MIG

Common Target File

Union Target File

  

TP

FP

FN

%Sens

%FDRa

TP

FP

FN

%Sens

%FDRa

TP

FP

FN

%Sens

%FDRa

95%CI

95%CI

95%CI

95%CI

95%CI

95 % CI

ACE

SNV

518

0

16

97.0

<0.1

706

1

22

97.0

0.1

562

2

30

94.9

0.4

95.2-98.3

<0.1-0.7

95.5-98.1

<0.1-0.8

(92.8-96.6)

(<0.1-1.3)

InDel

18

1

1

94.7

5.3

23

0

0

100

<0.1

37

0

3

92.5

<0.1

74.0-99.9

0.1-26.0

85.2-100

<0.1-14.8

(79.6-98.4)

(<0.1-9.5)

WGSb

SNV

499

0

35

93.4

<0.1

701

0

27

96.3

<0.1

573

0

19

96.8

<0.1

91.0-95.4

<0.1-0.7

94.6-97.5

<0.1-0.5

(95.0-98.1)

(0–0.6)

InDel

18

0

1

94.7

<0.1

23

0

0

100

<0.1

38

0

2

95.0

<0.1

74.0-99.9

<0.1-18.5

85.2-100

<0.1-14.8

(83.1-99.4)

(<0.1-9.3)

SSCR

SNV

504

1

30

94.4

0.2

684

4

44

94.0

0.6

545

2

47

92.1

0.4

92.1-96.2

<0.1-1.1

92.0-95.6

0.2-1.5

(89.6-94.1)

(<0.1-1.3)

InDel

17

1

2

89.5

5.6

21

1

2

91.3

4.5

37

0

3

92.5

<0.1

66.9-98.7

0.1-27.3

72.0-98.9

0.1-22.8

(79.6-98.4)

(<0.1-9.5)

SS

SNV

497

2

37

93.1

0.4

704

0

24

96.7

<0.1

562

1

30

94.9

0.2

90.6-95.1

<0.1-1.4

95.1-97.9

<0.1-0.5

(92.8-96.6)

(<0.1-1)

InDel

16

2

3

84.2

11.1

21

0

2

91.3

<0.1

37

0

3

92.5

<0.1

60.4-96.6

1.4-34.7

72.0-98.9

<0.1-16.1

(79.6-98.4)

(<0.1-9.5)

NX

SNV

465

1

69

87.1

0.2

650

1

78

89.3

0.2

484

0

108

81.8

<0.1

83.9-89.8

<0.1-1.2

86.8-91.4

<0.1-0.9

(78.4-84.8)

(<0.1-0.8)

InDel

19

0

0

100

<0.1

21

0

2

91.3

<0.1

31

1

9

77.5

3.1

82.4-100

<0.1-17.6

72.0-98.9

<0.1-16.1

(61.5-89.2)

(0.1-16.2)

NG

SNV

346

6

188

64.8

1.7

436

14

292

59.9

3.1

373

10

219

63.0

2.6

60.6-68.8

0.6-3.7

56.2-63.5

1.7-5.2

(59.0-66.9)

(1.3-4.7)

InDel

11

0

8

57.9

<0.1

11

1

12

47.8

8.3

20

1

20

50.0

4.8

33.5-79.7

<0.1-28.5

26.8-69.4

0.2-38.5

(33.8-66.2)

(0.1-23.8)

  1. FDR false discovery rate, FN false negatives, FP false positives, MIG medically interpretable genome, SENS Sensitivity, TP true positives aFDR is used in lieu of specificity due to a large skew in the TN, FP class distribution.
  2. aFDR is used in lieu of specificity due to a large skew in the TN, FP class distribution
  3. bIn WGS data, there was no difference in error rates when using either VQSLOD scores or hard-thresholding cutoffs for InDels