Nature - 15.08.2019

Article reSeArcH

the negative controls. Batch effects based on the use of particular pol-
ymerase chain reaction (PCR) reagent lots can also be visualized. For
example, the association of Thiohalocapsa halophila with either the
PCR reagent ‘5× Q5 buffer’ (lot 11408) or ‘Q5 Taq polymerase’ (lot

51405), both of which were used to process the same 390 samples, is

shown in Fig. 1f.

We used the kappa statistic to quantify the level of agreement

between 16S rRNA amplicon sequencing of two DNA samples from

Table 1 | Comparison of main signals using metagenomics with 16S rRNA amplicon sequencing

Species

Positive signals MG and 16S (79 = max) MG reads

in positive

samples

16S reads

in positive

samplesa

Concordance MG and 16S

kappa score

(P value)b

Part of an MG

batch effectc

Presence 16S in neg-
ative controls Absent/
Botha MG only 16S onlya Neither weak/strong (n = 5)d
Salmonella bongori 79 0 0 0 178 54% NA No 5/0/0
Escherichia coli 1 78 0 0 18,602 1.2% 0 (−) Gr. 1 & 2 4/1/0
Shigella (genus) 0 75 0 4 254 NA 0 (−) Gr. 1 & 2 5/0/0
Salmonella enterica 0 75 0 4 33 NA 0 (−) Gr. 1 & 2 5/0/0
Cronobacter sakazakii 0 65 0 14 21 NA 0 (−) Gr. 1 & 2 5/0/0
Bacillus subtilis 0 63 0 16 13 NA 0 (−) Gr. 1 & 2 5/0/0
Yersinia pseudotuberculosis 0 59 0 20 3 NA 0 (−) Gr. 1 & 2 5/0/0
Neisseria meningitidis 0 44 0 35 2 NA 0 (−) Gr. 1 & 2 5/0/0
Bradyrhizobium (genus) 0 79 0 0 125 NA 0 (−) Gr. 2 5/0/0
Rhodopseudomonas palustris 0 79 0 0 45 NA 0 (−) Gr. 2 5/0/0
Caulobacter (genus) 12 67 0 0 14 1.4% 0 (−) Gr. 2 1/3/1
Methylobacterium (genus) 9 69 0 1 8 2.4% 0.003 (0.36) Gr. 2 1/4/0
Burkholderia (genus) 21 57 0 1 7 1.9% 0.009 (0.27) Gr. 2 1/4/0
Propionibacterium acnes 66 13 0 0 20 4.8% 0 (−) No 0/3/2
Streptococcus pneumoniae 0 11 0 68 115 NA 0 (−) No 5/0/0
Vibrio cholerae 0 14 0 65 46 NA 0 (−) No 5/0/0
Thiohalocapsa halophila 0 0 71 8 NA 4.2% 0 (−) No 0/0/5
Stenotrophomonas maltophilia 5 51 1 22 2 1.9% 0.03 (0.24) No 2/3/0
Acinetobacter baumanii 1 26 0 52 2 2.4% 0.05 (0.08) No 4/1/0
Micrococcus luteus 1 46 0 32 15 2.0% 0.02 (0.20) No 4/1/0
Gardnerella vaginalis 0 5 0 74 1 NA 0 (−) No 4/1/0
Lactobacillus crispatus 0 4 0 75 1 NA 0 (−) No 5/0/0
Deinococcus geothermalis 1 1 0 77 68 33% 0.66 (<0.0001) No 5/0/0
Streptococcus agalactiae 3 4 0 72 8 13% 0.58 (<0.0001) No 5/0/0
The average number of metagenomics (MG) and average percentage of 16S reads in positive samples are shown. Gr., group; NA, not applicable.
a16S rRNA amplicon sequencing signals higher than 1% are defined as positive.
bOne-sided P values (kappa statistic).
cSee Fig.  1 for definition of groups 1 and 2.
dStrong signals are defined as more than 1%.

ab***

*** *

c *

Qiagen DNA isolation (%)

Mpbio DNA isolation (%)

Samples with more than 1% (%) Samples with more than 1% (%)

Vaginal lactobacilli

Vaginal lactobacilli

Vaginosis-associated

bacteria

Streptococcus

agalactiae

Vaginal (328)

Intrapartum (88)Pre-labour (82)

Vaginal (328)

Intrapartum (88)Pre-labour (82)

Vaginosis-associated bacteria

100

(^5030)
20
10
0
40
30
20
10
0
10
1
0.1
0.01
0.001
0.0001
0.0001 0.001 0.01 0.1 110 100
agaac obac
Vaginosis-associated
bacteria
gg
Streptococcus
agalactiae
Fig. 2 | Mode of delivery and detection of vaginal bacteria by 16S rRNA
amplicon sequencing. a, Concordant detection of vaginal lactobacilli and
a combination of all vaginosis-associated bacteria by both Qiagen (x axis)
and Mpbio (y axis) results in Spearman’s rho correlation coefficients of
0.37 and 0.59, respectively, when analysing the top right quadrant only
(>0.1%). b, c, Comparisons between vaginally associated bacteria and
the mode of delivery. *P < 0.05, ***P < 0.001, Mann–Whitney U-tests
were used where values below 1% are regarded as 0%. See Extended Data
Fig. 6 for scatterplots. Percentage read count is based on the higher value
for given species using Qiagen or Mpbio DNA isolation kit (using all 498
samples).
15 AUGUSt 2019 | VOl 572 | NAtUre | 331