KrakenUniq: confident and fast metagenomics classification using unique k-mer counts

Table 2 Performance of KrakenUniq (with unique k-mer count thresholds) compared to metagenomics classifiers [15] on the biological datasets (n = 10). F1 and recall show the average values over the datasets. Note that “KrakenUniq reads” would be equivalent to standard Kraken

	Genus		Species
	F1	Recall	F1	Recall	Avg
KrakenUniq nt k-mers	0.99	0.99	0.99	0.98	0.99
KrakenUniq nt reads	0.96	0.91	0.97	0.95	0.95
BlastMeganFilteredLiberal	0.97	0.94	0.97	0.89	0.94
BlastMeganFiltered	0.97	0.93	0.96	0.87	0.93
KrakenUniq orig k-mers	0.96	0.93	0.94	0.87	0.93
ClarkM4Spaced	0.95	0.90	0.94	0.88	0.92
KrakenUniq orig reads	0.95	0.90	0.94	0.85	0.91
Kraken	0.95	0.90	0.94	0.84	0.91
KrakenUniq std. k-mers	0.97	0.94	0.92	0.78	0.90
DiamondMegan_sensitive	0.98	0.93	0.92	0.74	0.89
KrakenFiltered	0.95	0.91	0.90	0.75	0.88
ClarkM1Default	0.94	0.85	0.91	0.77	0.87
KrakenUniq std. reads	0.95	0.89	0.90	0.70	0.86
LMAT	0.97	0.93	0.91	0.60	0.85
DiamondMegan	0.94	0.87	0.91	0.66	0.85
Gottcha	0.91	0.84	0.87	0.67	0.82
NBC	0.87	0.76	0.85	0.73	0.80
Metaphlan	0.94	0.89	0.83	0.55	0.80
MetaFlow	0.66	0.53	0.65	0.51	0.59
PhyloSift	0.68	0.29	0.78	0.54	0.57
PhyloSift90pct	0.68	0.30	0.77	0.52	0.57

ISSN: 1474-760X