Hybrid-hybrid correction of errors in long reads with HERO

Table 2 Benchmark results for correcting reads from real datasets. Indels/100 kbp: average number of insertion or deletion errors per 100,000 aligned bases. Mismatches/100 kbp = average number of mismatch errors per 100,000 aligned bases. Genome fraction (GF) reflects how much of each of the strain-specific genomes is covered by the corrected reads. N/100 kbp denotes the average number of uncalled bases (Ns) per 100,000 bases in the read. MC = fraction of misassembled contigs

Correction	Indels/100 kbp	Mismatches/100 kbp	GF (%)	N/100 kbp	MC (%)
Correct Bmock12 ONT reads
ONT reads	4001.58	2659.11	93.63	0.00	2.33
Ratatosk	31.17	94.91	93.74	0.67	2.64
R-HERO	6.45	40.48	93.78	0.20	2.57
FMLRC	8.09	21.82	93.77	0.00	2.93
F-HERO	5.87	20.82	93.77	0.06	2.92
LoRDEC	8.10	22.56	93.76	0.00	2.73
L-HERO	5.03	20.54	93.81	0.06	2.70
Correct Bmock12 PacBio reads
PacBio reads	7790.78	1851.43	87.98	0.00	7.55
Ratatosk	335.23	152.21	91.76	1.16	10.53
R-HERO	59.87	58.89	92.31	0.25	12.28
FMLRC	30.39	46.53	94.84	0.00	18.14
F-HERO	20.76	42.45	94.88	0.03	18.55
LoRDEC	40.99	60.44	93.00	0.00	15.32
L-HERO	18.95	53.01	93.51	0.04	17.09
Correct NWC ONT reads
ONT reads	7991.02	5476.30	100.00	0.00	7.18
Ratatosk	433.92	589.11	99.99	22.42	7.22
R-HERO	198.49	426.52	100.00	13.25	7.49
FMLRC	225.39	340.97	99.99	0.00	10.35
F-HERO	97.17	270.36	99.99	0.00	10.70
LoRDEC	154.4	489.82	99.99	0.00	7.18
L-HERO	72.05	377.92	99.99	0.00	9.65
Correct NWC PacBio reads
PacBio reads	11479.15	5270.85	81.61	0.00	7.54
Ratatosk	369.40	446.14	83.13	13.65	6.12
R-HERO	190.04	301.22	83.45	8.66	5.58
FMLRC	244.74	239.31	84.62	0.00	5.14
F-HERO	108.63	171.05	84.49	0.00	5.01
LoRDEC	79.42	422.12	83.19	0.00	5.01
L-HERO	54.68	373.06	83.00	0.00	4.76

Boldface is meant to indicate the best performing tool in the respective category

ISSN: 1474-760X