Article Figures & Data
Figures
- FIG 1
Compositional PCA plot of samples (A) and OTU loadings (B) for the initial data set. Only OTUs that had an absolute effect size difference between any two groups of ≥1 (24) or were compositionally associated with a ρ value of >0.65 (22) (Materials and Methods) are included in these plots. In panel A, each point is a sample and the distance between points is proportional to the multivariate difference between samples. Samples are colored by their age group membership, and the data ellipses encompass 75% of the points in a group (75% confidence interval [CI]). Panel B shows the loadings for panel A in the same coordinate space, which represent the contributions of the OTUs to the separation of the samples. In this plot, each point is an OTU (colored by its assigned taxonomic genus) and the distance and direction from the origin to the point representing an OTU is proportional to the standard deviation of that OTU in the data set. The distance between one OTU and another is inversely proportional to their compositional association: points that are close together may have concordant relative abundances across all samples. In comparing the two plots, we see, for example, that the 19- to 24-year-old age group (lower right quadrant in panel A) has a higher relative representation of Bacteroides (lower right quadrant in panel B). The ability to directly interpret the plot is limited by the proportion of variance explained (37.4% on the first component and 9.2% on the second component).
- FIG 2
Exploration of the data set with age as a continuous variable. The α diversity panel plots Shannon’s diversity (shdiv) across the age range (x axis). Each point is an individual sample, and the black line is the Loess line of best fit. Panels A through G represent clusters of concordant OTUs with an expected ρ value cutoff of >0.65; this metric provides a measure of the constancy of the ratio between OTUs and is a replacement for correlation (18, 22, 23). Each line in a cluster plot is the Loess line of best fit for the clr relative abundance (rAB on the y axis) of an individual OTU across age (x axis). A 0 value indicates that the relative abundance of an OTU is equal to the mean log2 relative abundance of all OTUs, while a positive or negative value indicates relative abundances greater or less than the mean log2 relative abundance, respectively. OTU lines of best fit are colored according to the genus that the OTU is classified into according to the key. Note that most of the clusters contain OTUs related by the same genus (A, C, D, F, and G). The lines of best fit suggest approximately equal ratios between the cluster members across the age range; however, this must be investigated further (22, 23), as shown in the last panel. For demonstration, the relative abundances between one pair of concordant OTUs from cluster C is plotted in the bottom right panel (OTU 6 versus OTU 3340). These two OTUs are the two relatively most abundant OTUs in cluster C (top two lines), and they have an expected ρ value of 0.8. The slope of association shown in the red line is 0.82. The blue line shows the ideal slope of 1 (22, 23). The Pearson correlation coefficient is 0.83. Table S2 contains the slope and correlation information for all pairwise correlated OTUs.
- FIG 3
Differential relative abundance of major taxa for successive pairwise comparisons. All possible comparisons of cohorts revealed 102 OTUs that were reproducibly different between at least one pair of cohorts. These were grouped into 26 classified genera and one set containing all unclassified OTUs. The plots show all OTUs in these 26 genera for pairwise comparisons between each successive age group pair: 3 to 6 versus 8 to 12, 8 to 12 versus 13 to 14, 13 to 14 versus 19 to 24, 19 to 24 versus 30 to 50, 30 to 50 versus 60 to 79, and 60 to 79 versus >94. Each comparison plot shows a point for each OTU binned by genus with the log2 standardized difference, the “effect” measure determined by ALDEx2 (24, 39), between the two groups on the x axis. Points are colored as red or blue if they have an effect size of ≥1 for the comparison. An effect size greater than 1 indicates that the OTU will be reliably found to have a greater difference between groups than dispersion within either group (24). Equivalent plots for comparisons at different taxonomic levels are shown in Fig. S6.
Tables
- TABLE 1
Summary of sample groups and sites
Group Age (yr) No. female, male Sampling location Children Kindergarten 3–6 60, 43 Dagang Central Kindergarten, Zhenjiang City, Jiangsu Province;
Taicang Weiyang Kindergarten, Suzhou City, Jiangsu Province;
Taicang Huasheng Kindergarten, Suzhou City, Jiangsu ProvincePrimary school 8–12 90, 71 Dagang Central Primary School, Zhenjiang City, Jiangsu Province;
Taicang Zhu Diwen Primary School, Suzhou City, Jiangsu ProvinceMiddle school 13–14 65, 49 Taicang No. 2 Middle School, Suzhou City, Jiangsu Province Adults Youth 19–24 55, 80 Taicang Chien-Shiung Institute of Technology, Suzhou City,
Jiangsu ProvinceMiddle age 30–50 66, 20 Taicang Xingda Can Co., Ltd., Suzhou City, Jiangsu Province Elderly 60–79 43, 43 Taicang Xinhu Community, Suzhou City, Jiangsu Province Centenarians >94 142, 56 RuGao City, Nantong City, Jiangsu Province Young soldiers 19–24 12, 200 Gansu police detachment training base and
PLA training base in Lanzhou, Gansu Province
FIG S1
PCA plot of the initial data set. These two plots show the first two principle components of a singular value decomposition of the centered log ratio-transformed data for the data set comprising samples from people aged 3 to ≥100 years. The left panel shows a plot where each point is a sample, where the distance between points is proportional to the difference between samples. Samples are colored by their grouping, and the 75% data ellipses demonstrate that groups separate broadly by age cohort. The right panel shows the contribution of the OTUs to the separation of the samples. The distance and direction from the origin to the point representing an OTU are proportional to the standard deviation of the OTU in the data set. The distance between two OTUs is inversely proportional to their compositional association: points that are close together may have concordant abundances across all samples. The ability to directly interpret the plot is limited by the small proportion of variance explained. OTUs are colored gray if deemed uninteresting, red if they were part of a group where E(ρ) is >0.65, blue if the effect size was ≥1 for any pairwise comparison between groups, or magenta if both approaches deemed them to be of further interest. See the Materials and Methods section for the method used to calculate these values. Download FIG S1, TIF file, 1.7 MB.
Copyright © 2017 Bian et al.
This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
FIG S2
Multivariate exploration of the data set. Panel A shows that samples from individuals that are near 20 years old separate from all other samples even when the young soldier samples are included in this data set; these are the 19- to 24S samples in the legend. As in Fig. 1, samples are filtered to include all taxa that occur with a minimum frequency in any sample of 0.001 and that occur in at least 20% of samples. Panel B shows an unsupervised clustering approach using the Aitchison distance (Euclidian distance of the clr values) and Ward.D2 clustering, including all OTUs. Samples are colored by group. Essentially, there are three large partitions. One was composed of the majority of members of the 3-to 6-year-old and 8-to-12-year-old groups, one composed of both 19- to 24-year-old groups, and one composed of the remainder. This last group is heterogeneous and likely results because the 13- to 14-year-old and 30- to 50-year-old groups are closest to the center of the data set. Panel C is a PCA plot of the entire data set, including only the OTUs that are different or compositionally associated. Samples from individuals that are in the 19- to 24-year-old groups are separate from all other samples. Panel D shows the loading plot for PCA in panel C and includes only the subset of OTUs that are different between groups with an effect size of ≥1 or that are compositionally associated. Download FIG S2, PDF file, 0.2 MB.
Copyright © 2017 Bian et al.
This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
TABLE S1
Table of PERMANOVA test of location values calculated using the Vegan Adonis function, using the Aitchison distance (method=euclidian) using clr-transformed values. The “R2” value is the proportion of the variance that is estimated to be caused by a location (difference in mean value) difference. “F.Model” represents the F statistic values, and “P” is the estimated PERMANOVA P value. Only the youth (“you”) and young soldier (“ys”) groups are strongly different from their bracketing cohorts. “MCD” is the median Aitchison distance of each sample to the group A centroid, and “IQR” is the interquartile range of those distances. The final row is a placeholder to show the MCD and IQR for the >94-year-old age group (cent). Values were calculated using the betadisper function in the vegan R package (40). The young soldier group is less disperse than the others, and there is a tendency for the two oldest groups, aged 60 to 79 (eld) and >94 (cent), to have greater dispersion. Download TABLE S1, DOCX file, 0.01 MB.
Copyright © 2017 Bian et al.
This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
FIG S3
Filtering by relative abundance does not change the conclusions. The left plot shows the data filtered to a minimum abundance in any sample of 1%, and the right plot shows the data filtered to a maximum abundance in any sample of 2%. This demonstrates that the split between groups is intrinsic to the composition of the microbiota as a whole and is not driven by a preponderance of rare or abundant OTUs in the groups. Download FIG S3, TIF file, 1.4 MB.
Copyright © 2017 Bian et al.
This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
FIG S4
Exploration of the data set suggests that sample group is likely the major explanatory factor. Download FIG S4, TIF file, 1.8 MB.
Copyright © 2017 Bian et al.
This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
FIG S5
Exploration of α diversity in the groups shows that diversity is correlated with the age gradient and that the read count (RC) is not associated with Shannon’s diversity (ShDiv). Download FIG S5, TIF file, 2.2 MB.
Copyright © 2017 Bian et al.
This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
TABLE S2
Summary of E(ρ) data. Each OTU pair has an expected value of ρ of >0.65; however, this measure is a single value that encapsulates both the slope and Pearson’s correlation of an association (18, 22). The “slope” and “corr” columns show the slope of the association and Pearson’s correlation coefficient. Slopes near 1 and correlations near 1 indicate a better fit to the model of compositional association. Download TABLE S2, DOCX file, 0.02 MB.
Copyright © 2017 Bian et al.
This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
FIG S6
Differential relative abundance of major taxa for successive pairwise comparisons grouped by family (top), class (middle), and phylum (bottom). All possible comparisons of group revealed many OTUs that were reproducibly different between at least one pair of cohorts. These were grouped into 26 genera and one set of unknown genera. Points are colored as red or blue if they have an effect size of ≥1 as determined by ALDEx2 for the comparison. Download FIG S6, TIF file, 2.8 MB.
Copyright © 2017 Bian et al.
This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
FIG S7
Examination of the entire data set using nonmetric multidimensional scaling (NMDS) shows that the ordinations are broadly similar in that the two most extreme groups are the samples from the youngest group and the 19- to 24-year-old group. The separation between the >94-year-old group and the others is less pronounced based on this method, which uses only rank and not relative abundance information. The relationship is not linear at low levels of dissimilarity. Download FIG S7, TIF file, 2.2 MB.
Copyright © 2017 Bian et al.
This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
TEXT S1
The exact workflow to generate the table used for 16S rRNA gene sequence analysis. Download TEXT S1, DOCX file, 0.02 MB.
Copyright © 2017 Bian et al.
This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.
