Skip to main content
. 2021 Oct 27;22:299. doi: 10.1186/s13059-021-02512-x

Table 3.

Benchmarking results for ONT data

Dataset Assembler Size (Mb) HC (%) k-mer recovery(%) Continuity (bp) QV Switch error(%) N (%) Dup (%)
All Mat Pat NG50 NGA50 Phased N50
MHC (ONT 25x) phasebook 14.5 97.0 98.4 94.0 87.7 152,599 152,599 102,954 36.9 0.78 0.0 1.82
phasebook-hi 14.8 91.6 97.5 89.4 88.8 796,423 746,361 176,895 40.6 6.99 0.0 1.96
Canu 5.9 59.6 95.4 83.2 65.9 1,419,330 93,217 100,918 33.4 6.44 0.0 0.84
Falcon 6.3 67.8 94.3 82.2 68.8 2,489,921 442,832 220,087 27.6 4.65 0.0 1.10
Flye 5.0 54.5 95.0 81.4 62.6 792,993 86,893 106,195 39.7 6.22 0.0 0.82
Shasta 5.0 63.0 86.8 62.5 48.5 957,250 136,119 147,280 23.7 10.71 0.0 0.89
Wtdbg2 5.0 76.5 89.1 67.0 47.4 4,821,843 208,838 136,335 24.3 4.33 0.0 0.99
HapCut2 9.1 55.9 84.5 55.6 52.9 493,723 279,218 367,400 31.5 6.33 8.9 1.53
WhatsHap 9.2 56.9 84.4 55.3 53.0 493,723 279,218 324,450 31.5 5.47 8.9 1.51
NA19240 (Chr6) (ONT 26x) phasebook 380.8 84.4 87.5 53.9 55.1 90,219 85,595 51,291 20.6 36.96 0.0 1.87
phasebook-hi 335.5 79.4 83.7 46.3 45.9 127,968 122,097 63,798 22.3 40.12 0.0 1.76
Canu 169.2 65.4 83.9 40.7 40.8 129,586 22.5 35.40 0.0 0.98
Falcon 179.3 64.0 77.7 33.7 35.4 69,660 37,993 162,461 20.5 35.72 0.0 1.02
Flye 167.4 57.3 89.4 40.1 48.3 115,043 25.1 31.85 0.0 0.99
Shasta 164.9 62.7 76.8 33.2 34.2 190,758 20.2 39.98 0.0 0.99
Wtdbg2 166.7 60.6 78.3 33.2 33.9 160,729 20.4 35.13 0.0 0.98
HapCut2 341.3 56.2 96.3 71.2 71.2 27,552,232 3,285,944 105,889 23.0 29.39 0.4 1.68
WhatsHap 341.3 56.4 96.2 71.1 71.3 27,552,232 3,520,811 102,096 23.0 28.45 0.4 1.70
HG002 (ONT 38x) phasebook 5691 92.0 62.2 63.2 390,659 202,723 26.6 2.28 0.0
Canu 2901 84.5 39.3 51.6 266,624 23.2 10.27 0.0
Flye 2928 83.9 38.5 50.6 287,619 23.0 11.94 0.0
Shasta 2805 88.8 40.9 53.2 276,802 25.0 12.59 0.0
Wtdbg2 2794 83.0 34.4 44.2 239,707 22.8 9.62 0.0

The publicly released assemblies of HG002 (Canu, Flye, Shasta, Wtdbg2) were directly used for comparison. We failed to run HapCut2 and WhatsHap for HG002 data on a computing machine (48 cores, 1 TB RAM) probably due to running out of memory