Skip to main content

Table 2 Evaluation of error correction performance

From: HALC: High throughput algorithm for long read error correction

Method

Throughput

Alignment ratio

Alignment identity

Genome fraction

N reads

Average read length

Sensitivity

Gain

Specificity

(a) Long reads of E. coli

Initial

100.0%

50.4%

95.2%

100.0%

75152

2381

-

-

-

PacBioToCAa

24.2%

100.0%

100.0%

99.5%

53447

810

-

-

-

LSC

53.5%

98.7%

99.9%

99.7%

115960

825

52.6%

51.7%

99.9%

Proovread

57.4%

100.0%

99.9%

99.7%

44986

2284

57.4%

56.8%

99.9%

CoLoRMap

42.8%

99.7%

100.0%

99.9%

70582

1084

42.7%

42.2%

99.9%

ECTools

23.5%

99.9%

99.2%

99.4%

8095

5211

23.4%

21.8%

99.8%

LoRDEC

60.8%

97.8%

100.0%

99.8%

70164

1549

60.7%

60.5%

100.0%

Jabba

52.8%

99.6%

100.0%

98.6%

26459

3568

52.8%

52.7%

100.0%

HALC

64.6%

98.6%

99.9%

99.8%

78731

1467

64.4%

64.0%

99.9%

(b) Long reads of A. thaliana

Initial

100.0%

32.4%

92.4%

82.4%

490418

2645

-

-

-

PacBioToCAa

10.7%

99.2%

99.7%

63.9%

260834

535

-

-

-

LSC

25.9%

100.0%

99.5%

71.4%

659123

509

24.2%

22.3%

99.7%

Proovread

27.8%

99.8%

99.7%

79.8%

125786

2864

26.5%

24.9%

99.7%

CoLoRMap

21.4%

99.4%

99.7%

69.3%

230933

1203

20.5%

19.2%

99.8%

ECTools

11.3%

99.8%

99.5%

63.1%

21354

6886

10.8%

9.8%

99.8%

LoRDEC

28.0%

86.4%

99.5%

74.4%

847963

428

25.9%

22.8%

99.6%

Jabba

10.8%

99.6%

99.7%

56.1%

51353

2726

10.5%

9.9%

99.9%

HALC

34.7%

96.5%

99.5%

85.8%

548872

819

33.2%

29.7%

99.3%

(c) Long reads of Maylandia zebra

Initial

100.0%

46.9%

91.3%

91.9%

1307812

10082

-

-

-

LoRDEC

33.6%

97.9%

99.7%

89.5%

7372455

601

32.4%

29.8%

99.6%

HALC

41.2%

98.7%

99.6%

90.7%

4833536

1123

40.2%

37.5%

99.4%

  1. The long reads of tests (a)-(c) are from E.coli, A. thaliana and Maylandia zebra, respectively. The initial and error corrected long reads by PacBioToCA, LSC, Proovread, CoLoRMap, ECTools, LoRDEC, Jabba and HALC are compared in the tests. The performance measurements are listed in the “Performance measurements” section.
  2. aSome measurements are not available without the correspondence information between a split long read and its initial long read