Skip to main content

Table 2 Performances (F1 scores in %) of SNP predictions in difficult-to-map regions and Major Histocompatibility Complex (MHC) by NanoCaller, Medaka, Clair, and Longshot on ONT data. These evaluations are performed against v4.2.1 benchmark variants for the Ashkenazim trio (HG002, HG003, and HG004), whereas “HG002 Bonito” and “HG002 R10.3” are different HG002 ONT datasets

From: NanoCaller for accurate detection of SNPs and indels in difficult-to-map regions from long-read sequencing by haplotype-aware deep neural networks

Prediction

Variant caller

HG002

HG003

HG004

HG002 Bonito

HG002 R10.3

SNPs on ONT data in difficult-to-map regions

NanoCaller ONT-HG001

95.80

96.83

96.70

97.44

96.34

NanoCaller ONT-HG002

96.18

96.92

96.92

97.38

96.44

Medaka

95.41

96.46

96.46

96.51

94.20

Clair

94.98

96.27

96.12

95.63

84.83

Longshot

93.95

94.61

94.55

95.42

93.00

SNPs on ONT data in MHC

NanoCaller ONT-HG001

98.65

99.06

99.18

99.45

98.46

NanoCaller ONT-HG002

98.86

99.19

99.28

99.46

98.69

Medaka

97.62

99.25

98.10

98.24

98.24

Clair

97.60

98.51

98.57

98.97

92.06

Longshot

68.52

73.13

69.40

68.48

68.41