NextDenovo: an efficient error correction and accurate assembly tool for noisy long reads

Table 1 Statistics of ONT read error correction

Source	Software	Corrected bases rate (%)	Average length (bp)	Max length (bp)	Reads with chimeric alignments (%)	Mapped with > = 99% coverage (%)	Mapped with > = 97% identity (%)	Average error rate (%)	Wall clock time (hour)
Simulation (chr1, 62X)	Raw reads	-	22,884	279,538	0.03	73.28	0.00	12.37	-
	Consent	84.72	23,597	280,569	0.03	97.01	99.77	0.14	7.29
	NextDenovo	83.47	23,515	188,302	0.03	99.58	99.96	0.20	2.43
	Necat	85.12	23,542	264,959	0.08	98.46	98.83	0.55	2.75
	Canu	81.26	23,369	279,129	0.41	98.03	96.58	2.02	18.08
CHM13 (chr1, 72X)	Raw reads	-	91,209	499,238	17.07	81.19	0.18	8.57	-
	Consent	99.18	90,701	503,237	19.22	80.72	83.11	1.37	17.43
	NextDenovo	97.13	90,981	505,469	10.70	89.10	89.31	0.90	1.83
	Necat	98.13	89,170	506,817	11.43	88.30	88.67	0.99	2.98
	Canu	92.59	84,270	502,469	13.44	85.13	80.64	2.21	126.72

Only the primary alignments defined by minimap2 of each read were used for evaluation. Corrected base rate is the ratio of the size of the corrected reads to the size of the raw reads to be corrected. Reads with chimeric alignments are defined as reads that have supplementary alignments. Average error rate only uses the reads that are mapped with ≥ 80% coverage. All the software was tested on the same computer with 32 CPUs and 252 GB RAM of memory. Best results for each metric are highlighted in bold

ISSN: 1474-760X