Skip to main content

Table 5 Time (T, in seconds) and recall (R, as a percentage of exact alignments) for aligners across different datasets on a single node

From: Fast noisy long read alignment with multi-level parallelism

Dataset

Size

Metric

Minimap2

Parasail

ParaHAT

WFA

WFA-GPU

(SSE)

(SSE)

(SSE+GPU)

(CPU)

(GPU)

E. coli

628

T (s)

69.37

110.43

46.70

3.10

2.45

R (%)

100

100

100

99.99

99.68

S. cer.

1569

T (s)

153.17

269.51

105.71

6.44

5.37

R (%)

100

99.93

100

100

99.29

D. mel. (sim)

10M

T (s)

795.21

1401.00

609.03

40.03

35.18

R (%)

100

99.99

100

99.89

97.97

A. tha.

10M

T (s)

823.94

1453.55

541.95

45.62

39.27

R (%)

100

99.99

100

100

99.07

H. sap. (sim)

10M

T (s)

795.17

1170.88

530.22

41.90

37.60

R (%)

100

99.99

100

99.97

98.60

H. sap. (real)

10M

T (s)

647.86

1158.48

477.44

23.13

20.87

R (%)

100

99.99

100

99.89

98.49

D. mel. (real)

10M

T (s)

767.07

1201.44

496.32

40.71

35.86

R (%)

100

99.99

100

99.99

97.99

PacBio CLR

10M

T (s)

1604.32

3031.9

820.56

29.98

25.25

R (%)

100

99.99

100

99.82

96.51

PacBio HiFi

10M

T (s)

2027.06

3822.86

1287.12

54.39

47.27

R (%)

100

99.99

100

99.98

98.53

ONT

10M

T (s)

3323.89

n/a

2108.81

66.69

45.64

R (%)

100

n/a

100

99.99

97.63

  1. \(^{1}\) All CPU executions use 16 threads. Executions taking more than 2 h are marked as n/a