Table 1 Accuracy of microbial identification using two in silico constructed simulated food mixtures.

From: Monitoring the microbiome for food safety and quality using deep shotgun sequencing

 

Mixture 1, no MF # GENERA

Mixture 1, MF # GENERA

Mixture 1, no MF # READS

Mixture 1, MF # READS

Mixture 2, no MF # GENERA

Mixture 2, MF # GENERA

Mixture 2, no MF # READS

Mixture 2, MF # READS

Bacteria in mixture (expected Content)

14

14

15,000

15,000

14

14

15,000

15,000

Observed bacteria

34

18

13,700

13,517

33

15

13,999

13,551

Observed viruses

9

0

563

0

4

0

328

0

Observed archaea

1

0

1

0

1

0

3

0

Observed Eukaryota

4

0

104

0

4

0

799

0

Total observed (TO)

48

18

14,368

13,517

42

15

15,129

13,551

True positives (TP)

14

14

13,571

13,511

14

14

13,623

13,548

TP as % of TO

29%

78%

94.45%

99.96%

33%

93%

90.05%

99.98%

False positives (FP)

34

4

797

6

28

1

1506

3

FP as % of TO

71%

22%

5.55%

0.04%

67%

7%

9.95%

0.02%

FP removed with MF

30

791

27

1503

% FP removed with MF

88.2%

99.2%

96.4%

99.8%

  1. The Simulated Food Mixtures (Mixture 1 and Mixture 2, see Supplementary Table 3) contain food matrix and microbial sequences. Microbial identification results are shown without matrix filtering (no MF) and with matrix filtering (MF). The number of observed genera (# GENERA) and observed genus-assigned reads (# READS) are shown for each category and summarized as the total observed (TO) counts. True positive (TP) and false-positive (FP) counts and fractions of TO are shown. The last two rows show the counts and percentages of false positives removed with matrix filtering.