]>

Table 4: MEGAN's top-ten strains for the Sargasso Sea dataset, their respective reads, and comparison to the NBC 9 mer and 15 mer methods. N/A means the strain is not in our training set (it is unfinished so it cannot be found. Burkholderia and Shewanella which were also found by Venter et al. [33] also have high matches in the NBC. The NBC's detection of Candidatus Pelagibacter drastically changes from to .

High strain content in sample (genome size—bothsides) MEGAN # of Reads NBC mer # of reads NBC mer # of reads Burkholderia 383 (9.3 M) 514 693 2044 Candidatus Pelagibacter ubique HTCC1062 (2.6 M) 323 13 111 Shewanella ANA-3 (10.3 M) 186 484 989 Procholorococcus marinus MIT 9312 (3.4 M) 125 28 24 Psychroflexus torquis ATCC 700755 (8.6 M) 119 N/A N/A Burkholderia cenoecapacia HI2424 (15.52 M) 102 106 219 Burholderia vietanamiensis G4 (16.8 M) 101 93 92 Burkholderia ambifaria/cepacia AMMD (15.06 M) 91 265 127 Shewanella oneidensis MR-1 (10.32 M) 78 79 297 Synechoccus sp. WH8102 (4.86 M) 75 68 82